AWS Transit Gateway Architecture Explained

AWS Transit Gateway and AWS Cloud WAN Architecture Explained

NETWORK INSIGHT · ENGINEERING GUIDE 20

AWS Transit Gateway
& AWS Cloud WAN
Architecture Explained

Modern enterprise networks no longer operate inside a single data centre or AWS Region. Applications span multiple VPCs, Regions, on-premises environments and distributed branches. Engineering these networks requires more than connectivity: it requires predictable routing, controlled segmentation, scalable transit, resilient hybrid integration and security inspection that preserves traffic symmetry.

This engineering guide explores the architecture behind AWS Transit Gateway and AWS Cloud WAN, from attachment routing and route-table propagation to global core network policies, inspection architectures and production troubleshooting.

TRANSIT GATEWAY CLOUD WAN VPC ROUTING NETWORK SEGMENTS DIRECT CONNECT AWS NETWORK FIREWALL MULTI-REGION BGP & HYBRID
Interactive Architecture Laboratory

AWS Global Network Topology

AWS Transit Gateway and Cloud WAN Architecture Explorer Educational network topology with application VPCs, inspection VPC, Transit Gateway, Cloud WAN regional core edges, hybrid connectivity and animated selected packet paths. AWS REGION A AWS REGION B Application VPC 10.10.0.0/16 Workload subnets Shared Services 10.20.0.0/16 DNS / Operations Transit Gateway Regional hub Inspection VPC Stateful firewall 10.30.0.0/16 Hybrid Edge DX / VPN BGP connectivity Regional Transit TGW / Core edge Regional VPC 10.40.0.0/16 On-Premises 172.16.0.0/16 Branch / Data Centre CONCEPTUAL TOPOLOGY · NOT AN AWS CONFIGURATION EXPORT
Transit / Core VPC Workloads Hybrid Connectivity Security Inspection
Traffic Path Explorer
Select an engineering scenario

Observe the logical path and inspect the corresponding routing decision.

Routing Decision Inspector
Control-plane interpretation
Simplified educational routing model. Actual results depend on attachments, associations, propagation, VPC routes, policy and supported AWS features.
01 / Architectural Context

Why AWS Transit Architecture Matters

An enterprise may begin with two VPCs and a simple peering connection. As the environment grows, however, application networks, shared services, security boundaries and hybrid connectivity create a much more complicated routing problem.

Point-to-point connectivity can become operationally difficult to manage. Engineers need to understand which networks may communicate, which transit route tables control forwarding, where security inspection occurs, and how regional or global connectivity is established.

AWS Transit Gateway provides a regional transit hub connecting supported attachments, including VPCs and VPNs. Its route tables control how traffic arriving from an attachment is forwarded toward reachable destinations.

AWS Cloud WAN introduces a policy-driven global core network. Core network edges operate in selected AWS Regions, and segments establish logical routing domains across the network. Connectivity and isolation are determined by the core network policy and the attachment configuration.

ENGINEERING DOMAIN 01
Connectivity

Design VPC attachments, regional transit, hybrid connectivity and multi-region communication without creating unnecessary point-to-point network relationships.

ENGINEERING DOMAIN 02
Routing Control

Understand route table association, route propagation, static routing, policy-based segmentation and the interaction between VPC routing and the transit network.

ENGINEERING DOMAIN 03
Security and Resilience

Engineer inspection boundaries, routing symmetry, segmentation, fault isolation, regional resilience and predictable failure recovery.

02 / Architecture Comparison

Transit Gateway vs AWS Cloud WAN

Both services can be part of enterprise AWS networking, but they expose different architectural abstractions. Understanding this distinction is essential before choosing a connectivity model.

Regional Transit Architecture

AWS Transit Gateway

A regional hub that connects supported network attachments and forwards traffic using Transit Gateway route tables.

  • Regional transit hub
  • Attachment-to-route-table associations
  • Static routes and supported propagation
  • Route-table-based segmentation
  • Inter-Region connectivity through TGW peering
  • Explicit VPC route-table integration
Global Policy-Driven Architecture

AWS Cloud WAN

A managed global core network that uses policies, segments and regional core network edges to coordinate connectivity.

  • Multi-Region core network
  • Core network policy configuration
  • Logical network segments
  • Segment sharing and isolation controls
  • Regional core network edges
  • Centralised network visibility and management
Critical Engineering Principle

Connectivity does not imply reachability. An attachment can exist and be operational while traffic still fails because the source route table, transit routing domain, destination route table, security policy or return path is incorrect.

03 / Engineering Outcomes

What You Will Engineer in This Guide

The following sections build from fundamental transit connectivity to advanced production investigation. Each major architecture concept is paired with an interactive engineering exercise.

01 Design scalable VPC transit topologies.
02 Investigate TGW route table associations and propagation.
03 Trace packet forwarding between VPCs.
04 Model Cloud WAN segments and core edges.
05 Build isolated routing and security domains.
06 Analyse stateful firewall inspection paths.
07 Explain Direct Connect, VPN and BGP integration.
08 Evaluate multi-region routing and resilience.
09 Troubleshoot real-world AWS network incidents.
10 Validate forwarding, segmentation and return-path consistency.
NETWORK INSIGHT · THE VISUAL AGE ENGINEERING GUIDE 20 / ANIMATED V2

Explore the AWS Transit Architecture

Navigate the engineering foundations of AWS Transit Gateway and AWS Cloud WAN — from regional VPC connectivity and route selection to global core networks, segmentation, firewall inspection, hybrid routing and production troubleshooting. Select any of the nine engineering sections below.

CLOUD WAN GLOBAL CORE TGW REGION A TGW REGION B VPC PRODUCTION VPC SHARED SERVICES
GLOBAL CORE · REGIONAL TRANSIT · SEGMENTED VPCS
ARCHITECTURE DOMAIN
Regional Transit + Global Cloud Networking
Transit Gateway, Cloud WAN core networks, VPC attachments and multi-region routing.
SECURITY DOMAIN
Segmentation + Stateful Inspection
Route-table isolation, Cloud WAN segments, inspection VPCs and AWS Network Firewall.
OPERATIONS DOMAIN
Hybrid Connectivity + Incident Response
Direct Connect, VPN, BGP, regional resilience and production troubleshooting.
SECTION 01
Transit Gateway Architecture Fundamentals

Regional hub-and-spoke networking, attachments, VPC routing and the AWS transit forwarding model.

EXPLORE ARCHITECTURE →
SECTION 02
Transit Gateway Route Tables and Routing Decisions

Attachment associations, route propagation, static routes, blackhole routes and route selection.

EXPLORE ROUTING →
SECTION 03
VPC Attachments and Inter-VPC Connectivity

Attachment subnets, Availability Zones, source and destination route tables, and complete packet paths.

EXPLORE PACKET PATHS →
SECTION 04
AWS Cloud WAN Core Network Architecture

Global networks, regional core edges, attachments, segments and centralised network policy.

EXPLORE GLOBAL CORE →
SECTION 05
Network Segmentation and Security Boundaries

Transit Gateway route-table isolation, Cloud WAN segments, shared services and controlled network reachability.

EXPLORE SEGMENTATION →
SECTION 06
Centralised Inspection and AWS Network Firewall

Inspection VPCs, Gateway Load Balancer, stateful firewall routing, appliance mode and service insertion.

EXPLORE INSPECTION →
SECTION 07
Hybrid Networking: Direct Connect and VPN

Enterprise WAN connectivity, Direct Connect gateways, BGP advertisements and VPN failover.

EXPLORE HYBRID ROUTING →
SECTION 08
Multi-Region Networking and Resilience

Transit Gateway peering, Cloud WAN regional edges, failure domains and recovery design.

EXPLORE RESILIENCE →
SECTION 09
Production Troubleshooting and Validation

Missing routes, policy failures, firewall asymmetry, hybrid incidents and systematic network fault isolation.

EXPLORE TROUBLESHOOTING →
Start the AWS Transit Engineering Journey Nine detailed sections · Architecture, routing, security, resilience and production operations
START SECTION 01 →
NETWORK INSIGHT · THE VISUAL AGE AWS TRANSIT GATEWAY + CLOUD WAN · ENGINEERING GUIDE 20
Engineering Guide 20 · Section 01

AWS Transit Gateway
Architecture Fundamentals

AWS Transit Gateway provides a regional network transit architecture that connects Amazon VPCs and supported hybrid network attachments through a central routing hub. It replaces many point-to-point connectivity relationships with an attachment-based forwarding model governed by Transit Gateway route tables.

Understanding Transit Gateway requires separating three concerns: network attachment, route-table control, and packet forwarding. A connected attachment is not automatically reachable from every other attachment. Reachability depends on routing configuration, segmentation and the return path.

Regional Transit Gateway operates as a regional networking resource.
Hub-and-Spoke Attachments connect networks to a shared transit routing domain.
Policy Control Route-table association and routes determine forwarding reachability.
01.1 / Regional Transit Design

From VPC Peering to a Transit Hub

VPC peering is useful for direct communication between two VPCs, but it does not provide transitive routing. A VPC cannot use an ordinary peering connection as a transit path to reach another peer.

As an enterprise expands to many application VPCs, shared-services VPCs and hybrid networks, a large collection of peering relationships can become operationally complex.

Transit Gateway introduces a central transit resource. Each participating VPC connects using a Transit Gateway VPC attachment. Traffic can then be forwarded between attachments when the relevant route tables permit it.

AWS REGION — REGIONAL TRANSIT DOMAIN AWS Transit Gateway Regional Transit Hub Application VPC 10.10.0.0/16 Shared Services 10.20.0.0/16 Security VPC 10.30.0.0/16 Hybrid Network VPN / Direct Connect Forwarding is controlled by attachment routing and Transit Gateway route tables.
Figure 1 — Conceptual Transit Gateway hub-and-spoke architecture. The animated path illustrates logical transit connectivity, not an automatically permitted route.

Engineering principle: Transit Gateway simplifies connectivity topology, but it does not eliminate routing design. Engineers must still configure VPC routes, Transit Gateway route tables, attachment associations and any required security controls.

01.2 / Architecture Components

The Three Fundamental Building Blocks

COMPONENT 01
Transit Gateway

The regional transit resource that forwards traffic between supported network attachments. It provides the central connectivity point for the transit architecture.

COMPONENT 02
Attachments

Logical connections between Transit Gateway and supported resources, including VPCs, VPN connections, Direct Connect gateway associations and Transit Gateway peering.

COMPONENT 03
Transit Route Tables

Routing domains that determine which destination attachment receives traffic arriving through an associated ingress attachment.

A Transit Gateway can have multiple route tables. Each attachment is associated with one Transit Gateway route table for forwarding decisions, while its routes may be propagated into multiple route tables where propagation is supported and enabled.

This separation is central to network segmentation. Production, development and shared-services attachments can use different ingress route tables, even though they connect to the same Transit Gateway.

01.3 / Data Plane

How a Packet Traverses Transit Gateway

Consider a packet sent from an application instance in 10.10.0.0/16 toward a service in 10.20.0.0/16.

The source subnet route table must direct the destination prefix toward Transit Gateway. The packet then enters the Transit Gateway through the source VPC attachment.

STEP 01 Source Workload Application generates traffic
STEP 02 VPC Route Table Destination points to TGW
STEP 03 Ingress Attachment Packet enters regional transit
STEP 04 TGW Route Lookup Associated route table evaluated
STEP 05 Destination VPC Forwarding and return routing

Transit Gateway performs a destination route lookup using the route table associated with the ingress attachment. If a usable route exists, the selected destination attachment receives the traffic.

The destination VPC must have the necessary subnet routing and security configuration. The reverse direction is evaluated independently, which makes return-path design particularly important.

ILLUSTRATIVE FORWARDING TRACE SOURCE: 10.10.1.25 DESTINATION: 10.20.2.40 [1] Application subnet Route: 10.20.0.0/16 → TGW [2] Source VPC attachment Packet enters Transit Gateway [3] Associated TGW route table Lookup: 10.20.0.0/16 Target: Shared Services attachment [4] Destination VPC attachment Forward to destination subnet [5] Return path 10.10.0.0/16 → TGW Reverse TGW lookup required RESULT: Connectivity succeeds only when forwarding, return routing and security policy are consistent.
01.4 / Control Plane

Route Association vs Route Propagation

These two mechanisms are frequently confused during Transit Gateway troubleshooting.

Mechanism Purpose Engineering Meaning
Association Selects the Transit Gateway route table used for traffic entering through an attachment. Determines the ingress forwarding domain.
Propagation Adds supported attachment routes into selected Transit Gateway route tables. Controls route visibility in those tables.
Static Route Explicitly defines a destination route and attachment target. Provides deliberate forwarding control.
Blackhole Route Discards traffic matching the selected destination. Can enforce explicit route-level blocking.

A useful mental model is that association determines which table is consulted, while propagation determines which supported routes appear in a table. Propagation does not, by itself, guarantee end-to-end reachability.

Important distinction: Transit Gateway route propagation does not automatically populate the subnet route tables inside attached VPCs. VPC route tables must be configured separately for traffic that should use the Transit Gateway.

01.5 / Availability Zone Design

VPC Attachment Subnets and AZ Reachability

When creating a Transit Gateway VPC attachment, engineers select attachment subnets in the required Availability Zones. A VPC attachment uses one selected subnet per Availability Zone.

The attachment subnet selection matters because Transit Gateway connectivity is enabled through the selected Availability Zones. Workloads in other Availability Zones may require additional attachment subnet coverage to use the transit connection.

For production architectures, engineers should evaluate multi-AZ attachment placement, subnet route tables, network ACLs, inspection placement and failure-domain resilience.

✓ Select appropriate attachment subnets across required AZs.
✓ Configure workload subnet routes toward Transit Gateway.
✓ Validate destination-side and return-path routing.
✓ Review stateful inspection symmetry and appliance mode.
✓ Verify security groups, NACLs and firewall policies.
✓ Test AZ failure and attachment reachability scenarios.
01.6 / Production Architecture

Engineering Principles for Scalable Transit

PRINCIPLE 01
Separate Routing Domains

Use dedicated Transit Gateway route tables to separate application environments, restricted networks and shared-services connectivity.

PRINCIPLE 02
Engineer Return Paths

Validate bidirectional routing. Stateful firewalls and inspection architectures require consistent traffic paths.

PRINCIPLE 03
Design for Failure

Consider Availability Zones, attachment states, regional dependencies, hybrid routing and operational observability.

The most reliable Transit Gateway deployments treat route tables as an explicit part of the network security architecture, not simply as a mechanism for connecting VPCs.

This foundation becomes essential when introducing centralised inspection, Direct Connect, VPN connectivity, Transit Gateway peering and Cloud WAN integration.

01.7 / Engineering Summary

The Transit Gateway Forwarding Model

AWS Transit Gateway centralises regional network connectivity, but forwarding remains dependent on explicitly engineered routing relationships.

The critical operational sequence is: source subnet route → ingress attachment → associated Transit Gateway route table → destination attachment → destination subnet → return path.

Next engineering concept: Section 2 examines Transit Gateway route tables in greater depth, including association, propagation, static routes, blackhole routes, route selection and segmentation failures.

NETWORK INSIGHT · THE VISUAL AGE Engineering Guide 20 · Editorial Section 1 of 9
Network Insight · Engineering Guide 20 · Lab 01

AWS Transit Gateway Architecture Explorer

Investigate the regional transit hub connecting application, database, and shared-services VPCs. Select source and destination networks, inspect attachment associations, trace route lookups, and introduce production-style failures.

Regional Transit Architecture

AWS Region: eu-west-1 · Example CIDRs

AWS REGION · EU-WEST-1 APPLICATION VPC 10.10.0.0/16 App subnet · 10.10.1.0/24 Attachment: tgw-att-app DATABASE VPC 10.20.0.0/16 DB subnet · 10.20.1.0/24 Attachment: tgw-att-db AWS TRANSIT GATEWAY Regional L3 Transit SHARED SERVICES VPC 10.30.0.0/16 DNS · Logging · Internal APIs
Transit hub Forwarding path Failed path Inactive path
3 VPC attachments
1 Regional TGW
3 / 3 Available attachments
Ready Packet path state
Select a VPC or Transit Gateway node to inspect its role in the architecture.

Traffic Engineering Console

Configure the source and destination, then run the simulation.
Forwarding Investigation
1. Source VPC route-table lookup
2. Transit Gateway attachment ingress
3. Associated TGW route-table lookup
4. Destination attachment and return path

Transit Gateway Control Plane

Example lab route table. Attachment association determines which TGW route table processes packets arriving from that attachment.

Destination Target Type Status

Lab simplification: all attachments initially associate with one TGW route table, with propagation enabled. Route-table isolation changes the source association in the model.

Attachment & Diagnostic Console

Attachment Association State
Simulated Diagnostic Output
[LAB READY] TGW topology loaded. Three VPC attachments available. Awaiting packet simulation.

Engineering Interpretation

AWS Transit Gateway is a regional Layer 3 transit service. VPC attachments provide connectivity into the hub, but connectivity also depends on VPC subnet route tables, Transit Gateway route-table associations, route entries, and return-path routing.

Educational model only. This widget does not query AWS, modify infrastructure, or perform live packet forwarding. Security groups, NACLs, DNS, AZ-specific attachment behavior, and stateful inspection are outside this first lab's forwarding model.

Engineering Guide 20 · Editorial Section 02

AWS Transit Gateway Route Tables and Routing Decisions

AWS Transit Gateway route tables define the forwarding behaviour of the regional transit network. They determine which destination attachments can be reached by traffic entering through a particular source attachment, providing both connectivity control and logical network segmentation.

A functioning Transit Gateway attachment does not guarantee connectivity. The decisive question is whether the route table associated with the ingress attachment contains a usable route toward the destination, and whether the surrounding VPC and hybrid networks have consistent forwarding and return paths.

02.1 / Routing Architecture

Transit Gateway Route Tables as Routing Domains

Each Transit Gateway route table represents an independent routing domain. Attachments can be associated with different route tables, allowing engineers to separate production, development, shared-services and restricted environments.

A Transit Gateway attachment is associated with one route table for ingress forwarding. Routes from eligible attachments may be propagated into one or more route tables, while static routes allow engineers to define explicit forwarding relationships.

AWS TRANSIT GATEWAY — LOGICAL ROUTING ARCHITECTURE Production VPC 10.10.0.0/16 Production attachment Development VPC 10.30.0.0/16 Development attachment Transit Gateway Ingress attachment Route table lookup Destination attachment Shared Services 10.20.0.0/16 Permitted destination Restricted VPC 10.40.0.0/16 Separate routing policy ASSOCIATION SELECTS INGRESS ROUTE TABLE · ROUTES SELECT DESTINATION ATTACHMENT
Figure 2 — Conceptual forwarding relationships between attachments and Transit Gateway routing domains. The highlighted path assumes the necessary routes and security permissions exist.

Engineering principle: Route-table association controls the forwarding context for traffic entering Transit Gateway. Route propagation controls route visibility in supported route tables. These are separate operations and should be validated independently.

02.2 / Routing Mechanisms

Association, Propagation, Static Routes and Blackholes

MECHANISM 01
Route Table Association

Determines which Transit Gateway route table evaluates traffic entering from an attachment. Each attachment has at most one active association.

MECHANISM 02
Route Propagation

Allows supported attachment routes to appear dynamically in selected Transit Gateway route tables. Propagation behaviour depends on attachment type and configuration.

MECHANISM 03
Static and Blackhole Routes

Static routes define explicit attachment targets. Blackhole routes deliberately discard traffic matching their destination prefix when selected.

VPC attachment route propagation is not equivalent to running BGP between a VPC and Transit Gateway. VPC CIDR information can be propagated through the AWS-managed attachment mechanism. BGP is relevant to supported VPN and Direct Connect routing arrangements.

Routing FunctionConfigurationOperational Effect
AssociationAttachment → TGW route tableSelects the ingress lookup domain.
PropagationEligible attachment → selected route tablesIntroduces supported reachable prefixes.
Static routePrefix → attachmentExplicitly determines the next attachment for a matching route.
Blackhole routePrefix → blackholeDiscards matching traffic when that route wins selection.
VPC subnet routePrefix → TGWDirects workload traffic into the transit network.
02.3 / Forwarding Decision Pipeline

How Transit Gateway Selects a Destination Attachment

The forwarding decision begins with the source VPC subnet route table, not the Transit Gateway route table. The VPC must first direct traffic to Transit Gateway. Once traffic enters through an attachment, Transit Gateway evaluates the associated route table.

01Source VPC RouteDestination points toward TGW
02Ingress AttachmentPacket enters transit
03Associated TGW TableLookup domain selected
04Route SelectionBest matching route evaluated
05Egress AttachmentPacket forwarded or discarded

Transit Gateway uses longest-prefix matching. When routes have the same prefix length, AWS-defined route priorities and route types affect selection. A static route generally takes precedence over a propagated route for the same destination prefix; AWS also defines attachment-specific priorities for competing propagated routes.

ILLUSTRATIVE ROUTE TABLE: PRODUCTION-RT Destination Target Type ------------------------------------------------------ 10.10.0.0/16 Production attachment Propagated 10.20.0.0/16 Shared Services Propagated 10.30.0.0/16 No route - 10.40.0.0/16 Blackhole Static 172.16.0.0/16 Hybrid attachment Propagated PACKET: Source: 10.10.1.25 Destination: 10.20.2.40 Ingress: Production attachment Lookup: Production-RT Match: 10.20.0.0/16 Target: Shared Services attachment RESULT: FORWARD NOTE: Return routing and security controls must also permit the communication.

Important: A blackhole route is not a universal deny rule. A more-specific matching route can win longest-prefix selection. For security isolation, route tables should be designed carefully and combined with appropriate security controls.

02.4 / Network Segmentation

Using Multiple Transit Gateway Route Tables

Network segmentation is achieved by associating different groups of attachments with separate route tables and controlling the destination routes available in each table.

For example, production and development attachments can each access a shared-services VPC while remaining isolated from one another. This requires carefully designed forwarding and return-path routing. The shared-services attachment's associated route table must contain the appropriate return routes.

Ingress Routing DomainPermitted DestinationsIsolated Destinations
Production-RTProduction, Shared Services, approved HybridDevelopment, Restricted
Development-RTDevelopment, Shared ServicesProduction, Restricted
Shared-RTApproved Production and Development prefixesRestricted
Restricted-RTRestricted and explicitly approved inspection pathsAll unapproved networks

Transit Gateway segmentation operates at the routing layer. It does not replace security groups, network ACLs, firewall inspection, identity controls or application-level authorisation.

02.5 / Route Selection Engineering

Longest Prefix, Route Priority and Failure Behaviour

Consider a Transit Gateway route table containing a broad route to a hybrid network and a more-specific route to a different attachment. The more-specific prefix is selected for matching traffic, regardless of the broader route's type.

EXAMPLE: LONGEST-PREFIX MATCH 10.0.0.0/8 → Hybrid attachment 10.20.0.0/16 → Shared Services attachment 10.20.50.0/24 → Blackhole Destination: 10.20.50.25 Selected: 10.20.50.0/24 Result: DROP Destination: 10.20.60.25 Selected: 10.20.0.0/16 Result: FORWARD to Shared Services Destination: 10.80.1.10 Selected: 10.0.0.0/8 Result: FORWARD to Hybrid attachment

Operational failures may occur when propagated routes disappear, static routes target unavailable attachments, attachment associations change or overlapping prefixes create unexpected forwarding decisions.

FAILURE 01
Missing Route

The ingress route table has no matching destination prefix. Traffic cannot be forwarded toward the required attachment.

FAILURE 02
Wrong Association

The source attachment uses a different route table than expected, causing the packet to follow an unintended routing policy.

FAILURE 03
Asymmetric Routing

Forward and reverse traffic take different inspection paths, potentially breaking stateful firewall sessions.

02.6 / Production Troubleshooting

Diagnosing Transit Gateway Routing Failures

Troubleshooting should follow the packet's forwarding sequence rather than assuming that an available attachment implies connectivity.

✓ Confirm the source subnet route points toward the correct Transit Gateway.
✓ Verify the VPC attachment state and Availability Zone coverage.
✓ Identify the route table associated with the ingress attachment.
✓ Inspect the effective matching route and its attachment target.
✓ Check propagation settings and static route configuration.
✓ Confirm destination subnet and reverse-direction routing.
✓ Review security groups, NACLs and inspection firewall policy.
✓ Use Transit Gateway Flow Logs, VPC Flow Logs and relevant reachability analysis tools where supported.

Production investigation rule: Trace both directions of a flow. A correct outbound route does not prove that the destination can return traffic, and successful routing does not prove that a firewall or security policy will permit the session.

02.7 / Engineering Summary

Routing Policy Defines Transit Reachability

Transit Gateway route tables provide the central forwarding and segmentation mechanism for regional AWS transit architectures. Attachment association chooses the ingress routing domain, propagation and static configuration populate route entries, and destination-prefix matching determines the selected route.

The next architectural layer is the VPC attachment itself: subnet selection, Availability Zone enablement, workload route tables and bidirectional packet forwarding.

Next section: AWS VPC Attachments and Inter-VPC Connectivity — attachment subnets, Availability Zone behaviour, packet paths, routing symmetry and troubleshooting failed communication between VPCs.

Network Insight · Engineering Guide 20 · Lab 02

Transit Gateway Route Table Simulator

Engineer routing decisions across three VPC attachments, two Transit Gateway route tables, propagated routes, static entries and blackhole policies. Change the control plane, then trace how packets are forwarded or dropped.

Interactive TGW Control-Plane Architecture

Select an attachment to inspect its association. Animated links highlight the selected forwarding path.

AWS REGION · EU-WEST-1 APPLICATION VPC 10.10.0.0/16 tgw-att-app DATABASE VPC 10.20.0.0/16 tgw-att-db TRANSIT GATEWAY Route Lookup SHARED SERVICES VPC 10.30.0.0/16 tgw-att-shared
TGW Selected path Drop Other links
3VPC attachments
2TGW route tables
3 Active routes in selected table
Select a network component to inspect its role.

Routing Engineering Console

Configure the ingress attachment and destination IP, then execute a route lookup.
Forwarding Decision Pipeline
1. Identify ingress attachment
2. Resolve associated TGW route table
3. Perform longest-prefix lookup
4. Forward or discard packet

Route Table Associations & Propagation

Change the route table associated with each attachment. Toggle propagation to control which propagated routes appear in each table.

Attachment Association Main Prop. Isolated Prop.
Engineering rule
Association determines the ingress lookup table. Propagation determines whether eligible routes from an attachment are advertised into a particular TGW route table. They are independent controls.

Live TGW Route Table

Viewing tgw-rt-main

Destination Target Type Action
Add Static or Blackhole Route
Static routes are installed in the currently selected route table.

Simulated Route Lookup Diagnostics

[NETWORK INSIGHT · TGW LAB 02] Control plane initialized. Default association: tgw-rt-main Route propagation: enabled Awaiting lookup.

Engineering Interpretation

Association: Selects the route table used for ingress traffic.
Propagation: Populates eligible attachment routes.
Static routing: Allows explicit attachment targets or blackholes.
Longest prefix: The most specific matching destination route wins.

Production Engineering Notes

AWS Transit Gateway routing is controlled by attachment associations, route propagation, static routes, and blackhole entries. A packet arriving through an attachment is evaluated against the TGW route table associated with that ingress attachment.

This laboratory models TGW route selection, not complete VPC forwarding. Actual connectivity additionally requires correct VPC subnet routes, destination attachment subnet coverage, security groups, network ACLs, and return paths. Static routes take precedence over propagated routes for identical prefixes; more-specific prefixes take precedence over less-specific ones.

Educational simulation: no AWS API calls or live network changes. This model uses VPC attachments, IPv4 prefixes, deterministic route selection and simplified route propagation. It does not model equal-prefix propagated-route priority across attachment types, appliance mode, or all AWS control-plane constraints.

Engineering Guide 20 · Editorial Section 03

AWS VPC Attachments and Inter-VPC Connectivity

Transit Gateway VPC attachments connect individual Amazon VPCs to the regional transit network. They provide the attachment interfaces through which packets enter and leave Transit Gateway, but successful communication requires coordinated subnet routing, attachment Availability Zone coverage, Transit Gateway route tables and return-path configuration.

A VPC attachment is not simply a logical checkbox enabling communication. Its subnet placement, route-table integration and forwarding behaviour are critical elements of production network design.

03.1 / Attachment Architecture

How VPCs Connect to Transit Gateway

A Transit Gateway VPC attachment connects one VPC to a Transit Gateway in the same AWS Region. During attachment creation, engineers select subnets from the Availability Zones that require transit connectivity. Only one subnet per Availability Zone can be selected for a particular VPC attachment.

These selected subnets provide the Transit Gateway attachment interfaces. Workload subnets do not have to be the same subnets used for attachment, but their route tables must direct the appropriate destination prefixes toward Transit Gateway.

The architecture therefore separates workload subnet placement, attachment subnet placement, and transit forwarding policy.

AWS REGION — MULTI-AZ TRANSIT GATEWAY CONNECTIVITY APPLICATION VPC — 10.10.0.0/16 AVAILABILITY ZONE A Workload 10.10.1.0/24 TGW Subnet 10.10.10.0/28 AVAILABILITY ZONE B Workload 10.10.2.0/24 TGW Subnet 10.10.20.0/28 Transit Gateway Regional hub SHARED SERVICES VPC — 10.20.0.0/16 AVAILABILITY ZONE A TGW Subnet 10.20.10.0/28 Services 10.20.1.0/24 AVAILABILITY ZONE B TGW Subnet 10.20.20.0/28 Services 10.20.2.0/24 CYAN: FORWARD FLOW · GOLD: ILLUSTRATIVE REVERSE FLOW
Figure 3 — Conceptual multi-AZ VPC attachment architecture. Actual Transit Gateway forwarding is AWS-managed; paths are simplified to illustrate attachment subnet coverage and bidirectional connectivity.

Engineering principle: An attachment subnet enables Transit Gateway connectivity for its Availability Zone. Workload subnet routes, attachment subnet route tables, Transit Gateway routing and security policy must all be evaluated when investigating connectivity.

03.2 / Subnet and Route Table Integration

Workload Subnets vs Attachment Subnets

A common misconception is that every workload subnet must be selected during Transit Gateway attachment creation. Instead, the attachment uses selected subnets to establish connectivity for the corresponding Availability Zones.

Workload subnet route tables direct packets toward Transit Gateway. Attachment subnet route tables must also support the intended forwarding paths within the VPC, particularly when traffic arrives from the transit network and must reach a destination subnet or inspection appliance.

SUBNET ROLE 01
Workload Subnet

Hosts application instances and services. Its route table determines whether traffic destined for another network is sent toward Transit Gateway.

SUBNET ROLE 02
Attachment Subnet

Contains the Transit Gateway attachment interface for an enabled Availability Zone. Its routing configuration influences traffic entering the VPC from Transit Gateway.

SUBNET ROLE 03
Transit Routing Domain

The Transit Gateway route table associated with the ingress attachment determines the destination attachment selected for forwarding.

Routing LocationExample DestinationTargetPurpose
Application workload subnet10.20.0.0/16Transit GatewaySend remote VPC traffic into transit.
Production TGW route table10.20.0.0/16Shared Services attachmentSelect destination VPC.
Shared Services attachment subnet10.20.0.0/16Local VPC routeReach destination services inside the VPC.
Shared Services workload subnet10.10.0.0/16Transit GatewayProvide return routing.
Shared Services TGW ingress table10.10.0.0/16Application attachmentComplete reverse transit forwarding.
03.3 / Inter-VPC Packet Forwarding

Tracing a Packet Between Two VPCs

Consider an application instance at 10.10.1.25 communicating with a shared service at 10.20.1.40. Both VPCs are attached to the same Transit Gateway, and their routing domains permit communication.

01Source Workload10.10.1.25 initiates traffic
02Source VPC Route10.20.0.0/16 → TGW
03Transit GatewayIngress route-table lookup
04Destination AttachmentForward toward Shared Services
05Destination Workload10.20.1.40 receives packet
INTER-VPC FORWARDING TRACE SOURCE: 10.10.1.25 DESTINATION: 10.20.1.40 APPLICATION VPC: Route: 10.20.0.0/16 → TGW TRANSIT GATEWAY: Ingress: Application attachment Associated RT: Production-RT Match: 10.20.0.0/16 Target: Shared Services attachment SHARED SERVICES VPC: Destination subnet: 10.20.1.0/24 Local VPC routing delivers traffic RETURN: Source: 10.20.1.40 Destination: 10.10.1.25 Route: 10.10.0.0/16 → TGW TGW reverse lookup → Application attachment RESULT: Bidirectional connectivity when routing and security policies permit.

A successful forward path does not prove that a session can be established. The return path must independently pass its VPC routing, Transit Gateway route-table lookup and security controls.

03.4 / Availability Zone Behaviour

Multi-AZ Attachment Placement and Resilience

Transit Gateway VPC attachment connectivity is enabled through the selected attachment subnets. Selecting attachment subnets in multiple Availability Zones provides connectivity coverage for workloads distributed across those zones.

An engineer should not assume that an attachment subnet selected in one Availability Zone automatically enables equivalent transit access for workloads in every other Availability Zone. Transit Gateway's VPC attachment AZ behaviour must be considered explicitly.

Production designs typically select attachment subnets across the Availability Zones hosting workloads that require transit connectivity. Routing, security policies and appliance placement must then be validated for each relevant zone.

✓ Enable Transit Gateway attachment coverage in required Availability Zones.
✓ Use appropriate attachment subnet sizing and placement.
✓ Validate route tables associated with workload and attachment subnets.
✓ Check NACL rules affecting attachment and workload subnet traffic.
✓ Evaluate inspection appliance symmetry across Availability Zones.
✓ Test connectivity during attachment or AZ disruption.

Production design rule: Multi-AZ connectivity is not merely a high-availability checkbox. Attachment subnet selection, routing policy and inspection design determine which failure scenarios the architecture can tolerate.

03.5 / Security and Routing Symmetry

Why Return Paths Matter

Inter-VPC routing can be technically correct while applications still fail because of security group rules, network ACLs, host firewalls or stateful inspection.

When a central inspection VPC is introduced, traffic may be deliberately steered through firewall appliances. The return path must traverse the appropriate stateful inspection context. Transit Gateway appliance mode can help maintain flow symmetry for supported appliance VPC attachment architectures, but it does not automatically correct every routing or firewall design error.

Failure ConditionLikely SymptomEngineering Investigation
Missing source VPC routeTraffic never reaches TGW.Inspect source subnet route table.
Incorrect TGW associationUnexpected destination lookup or no route.Verify ingress attachment association.
Missing destination routePacket cannot reach the destination attachment.Inspect associated TGW route table.
Missing return routeOne-way traffic or session timeout.Trace reverse VPC and TGW forwarding.
Security group or NACL denialTraffic dropped despite valid routes.Inspect security policy and flow logs.
Asymmetric inspectionStateful firewall drops return traffic.Check appliance mode, AZ selection and inspection routes.
03.6 / Production Troubleshooting

Diagnosing Failed Inter-VPC Connectivity

A disciplined investigation begins at the source workload and follows the complete forwarding chain. Engineers should avoid starting with Transit Gateway route tables before confirming that the source VPC is actually sending traffic to Transit Gateway.

PRODUCTION TROUBLESHOOTING CHECKLIST [1] SOURCE INSTANCE Confirm destination IP and application port. Check host firewall and network interface. [2] SOURCE SUBNET ROUTE TABLE Verify destination prefix → Transit Gateway. [3] SOURCE VPC ATTACHMENT Confirm attachment is available. Validate required AZ coverage. [4] TRANSIT GATEWAY ROUTING Identify ingress-associated route table. Inspect matching destination route. Confirm target attachment state. [5] DESTINATION VPC Validate attachment subnet routing. Confirm destination subnet reachability. Check security groups and NACLs. [6] REVERSE PATH Trace destination → source routing. Verify reverse TGW lookup. [7] OBSERVABILITY Review VPC Flow Logs. Review Transit Gateway Flow Logs. Use Reachability Analyzer where supported. Correlate with firewall and host logs.

VPC Flow Logs and Transit Gateway Flow Logs provide useful evidence, but they should be interpreted with awareness of their visibility and collection limitations. Flow-log acceptance alone does not prove that an application successfully completed a session.

03.7 / Engineering Summary

Attachments Are the Foundation of Transit Connectivity

AWS Transit Gateway VPC attachments connect workload networks to the regional transit architecture. Their Availability Zone placement determines connectivity coverage, while VPC route tables and Transit Gateway route tables jointly determine forwarding.

The essential engineering sequence is: workload subnet route → attachment AZ connectivity → Transit Gateway route lookup → destination attachment → destination subnet → reverse forwarding path.

Next section: AWS Cloud WAN Core Network Architecture — global networks, core network edges, policy-driven routing, segments, attachments and multi-region connectivity.

Network Insight · Engineering Guide 20 · Lab 03

AWS VPC Packet Path Explorer

Trace packets from an EC2 application workload, through a VPC subnet route table and Transit Gateway attachment, into the regional transit fabric, and onward to a database VPC. Validate the return path and investigate Availability Zone, routing, and security failures.

End-to-End AWS Packet Forwarding Architecture

The diagram shows two VPCs, regional TGW connectivity, attachment subnets, and two Availability Zones. Select components to inspect their forwarding responsibilities.

AWS REGION · EU-WEST-1 APPLICATION VPC 10.10.0.0/16 DATABASE VPC 10.20.0.0/16 AVAILABILITY ZONE A AVAILABILITY ZONE B AVAILABILITY ZONE A AVAILABILITY ZONE B EC2 APPLICATION 10.10.1.10 · 10.10.1.0/24 EC2 APPLICATION 10.10.2.10 · 10.10.2.0/24 DATABASE WORKLOAD 10.20.1.10 · 10.20.1.0/24 DATABASE WORKLOAD 10.20.2.10 · 10.20.2.0/24 TRANSIT GATEWAY L3 Forwarding tgw-att-app tgw-att-db Logical forwarding diagram · Not physical topology
Application Database Forwarding Failure
2VPC attachments
2Availability Zones
Ready Connectivity status
Select an application subnet, database subnet, or Transit Gateway to inspect its routing role.

Packet Path Engineering Console

Network Control Plane
Source VPC route to TGW
TGW destination route
Destination VPC return route
Source AZ TGW attachment coverage
Database security group allows traffic
Network ACL allows bidirectional traffic
Configure the network and execute the packet trace.

VPC and Transit Gateway Route Tables

Inspect the logical forwarding entries required for application-to-database connectivity.

Route table Destination Target Status

Routes are simplified to the prefixes relevant to this exercise. Real VPC route tables also contain local routes and may include more-specific routes, endpoints, NAT gateways, and other targets.

Packet Forwarding Pipeline

1. Source subnet route-table lookup
2. Source AZ attachment availability
3. TGW ingress and route-table lookup
4. Destination VPC forwarding
5. Security group and NACL evaluation
6. Return-path validation

Production Diagnostic Console

[NETWORK INSIGHT · VPC PACKET PATH LAB] AWS Region: eu-west-1 Application VPC: 10.10.0.0/16 Database VPC: 10.20.0.0/16 Transit Gateway: tgw-core Status: READY Awaiting packet trace.

Engineering Investigation

Source VPC: The subnet route table must direct the destination prefix to Transit Gateway.
Transit Gateway: The ingress attachment's associated route table must contain a valid destination route.
Destination VPC: Routing and security controls must permit traffic to reach the workload.
Return path: Bidirectional communication requires a valid route back to the source.

Production Architecture: Important AWS Constraints

A Transit Gateway VPC attachment uses selected subnets, with at most one attachment subnet per Availability Zone. Attachment subnet selection controls the Availability Zones through which the TGW can forward traffic for that VPC. Workload subnet route tables must direct remote destinations to the TGW, and attachment subnet routing must also be considered.

A workload in an Availability Zone without an attachment subnet may not be able to send traffic to the TGW through that VPC attachment. Availability Zone identifiers can differ between AWS accounts, so AZ IDs are more reliable than AZ names when coordinating multi-account designs.

Security groups are stateful, whereas network ACLs are stateless. Network ACLs must permit the relevant forward and return traffic, including ephemeral ports where applicable.

This is an educational simulation, not a live AWS network emulator. The diagram abstracts the internal TGW data plane, attachment ENIs, AZ-specific destination selection, and security processing. It does not query AWS or change infrastructure.

Engineering Guide 20 · Editorial Section 04

AWS Cloud WAN Core Network Architecture

AWS Cloud WAN provides a managed global network architecture for connecting and segmenting enterprise networks across multiple AWS Regions, VPCs, branches and hybrid environments. Instead of configuring every regional routing relationship independently, engineers define a core network policy that controls network segments, attachment relationships and connectivity behaviour.

The central architectural components are global networks, core networks, core network edges, network segments and attachments. Together, these provide a policy-driven approach to multi-region connectivity.

04.1 / Global Network Architecture

From Regional Transit to a Global Core Network

AWS Transit Gateway operates as a regional transit resource. Multi-region architectures using Transit Gateway commonly require explicit inter-region peering and coordinated routing configuration.

AWS Cloud WAN introduces a global core network abstraction. Engineers specify the AWS Regions in which core network edges are deployed and define a core network policy describing the intended network segmentation and connectivity.

The Cloud WAN service manages the underlying core network infrastructure according to the deployed policy. Attachments connect supported networks to the regional core network edges.

AWS CLOUD WAN — GLOBAL CORE NETWORK REGION A REGION B REGION C Core Network Edge A Regional edge Core Network Edge B Regional edge Core Network Edge C Regional edge Production VPC 10.10.0.0/16 Development VPC 10.30.0.0/16 Shared Services 10.20.0.0/16 Development VPC 10.35.0.0/16 Production VPC 10.40.0.0/16 Hybrid Network Branch / On-Premises Core Network Policy Segments · Attachment Rules · Routing Controls
Figure 4 — Conceptual AWS Cloud WAN architecture spanning three AWS Regions. Production and development routing domains are represented by distinct traffic colours. Actual connectivity depends on the deployed core network policy.

Engineering principle: AWS Cloud WAN does not automatically make every connected VPC reachable from every other VPC. Segment membership, attachment configuration, routing policy and security controls determine which paths are available.

04.2 / Core Network Components

The Five Building Blocks of AWS Cloud WAN

COMPONENT 01
Global Network

A global network provides an organisational and monitoring framework for the enterprise network, including Cloud WAN core networks and supported network resources.

COMPONENT 02
Core Network

The managed connectivity infrastructure that spans configured AWS Regions and applies the deployed core network policy.

COMPONENT 03
Core Network Edge

A regional component of the core network that provides connectivity for supported attachments in the corresponding Region.

COMPONENT 04
Network Segment

A logical routing domain used to separate networks and define permitted connectivity relationships.

COMPONENT 05
Core Network Policy

The declarative configuration describing regional edges, segments, attachment policies and supported routing controls.

COMPONENT 06
Attachments

Connections between the core network and supported resources such as VPCs, VPNs, Connect attachments and Transit Gateway route table attachments.

04.3 / Network Segmentation

Segments as Global Routing Domains

Network segments provide a central mechanism for logical isolation. An enterprise can define segments such as Production, Development, Shared Services and Restricted.

Attachments are associated with segments according to the core network policy. By default, separate segments are isolated, unless supported policy mechanisms explicitly permit connectivity between them.

Segment sharing can allow selected routing domains to exchange connectivity, but it should be designed carefully to avoid unintended network reachability.

Segment Typical Attachments Connectivity Objective
Production Production application VPCs Controlled multi-region production connectivity.
Development Development and testing VPCs Isolated non-production routing domain.
Shared Services DNS, identity and operations VPCs Approved access from selected segments.
Restricted Regulated or sensitive workloads Strictly controlled connectivity and inspection.

Security architecture: Routing segmentation limits network reachability, but it does not replace AWS security groups, network ACLs, firewall inspection or application authorisation.

04.4 / Declarative Control Plane

How Core Network Policies Define Connectivity

Cloud WAN uses a declarative core network policy to define the intended network architecture. Engineers can specify regional edge locations, segments and attachment policies rather than manually configuring every individual regional transit relationship.

Policy changes are evaluated and deployed through the Cloud WAN policy workflow. Operational validation should confirm that the deployed policy version matches the intended architecture.

ILLUSTRATIVE CLOUD WAN CORE NETWORK POLICY { "version": "2021.12", "core-network-configuration": { "asn-ranges": ["64512-64520"], "edge-locations": [ {"location": "eu-west-1"}, {"location": "us-east-1"}, {"location": "ap-southeast-1"} ] }, "segments": [ {"name": "production"}, {"name": "development"}, {"name": "shared-services"} ], "attachment-policies": [ { "rule-number": 100, "condition-logic": "or", "conditions": [ { "type": "tag-value", "key": "Segment", "operator": "equals", "value": "production" } ], "action": { "association-method": "constant", "segment": "production" } } ] } NOTE: This is a simplified policy example. Additional attachment rules, security controls and deployment validation are required for production use.

In production, engineers should manage policy versions through infrastructure-as-code workflows, controlled change reviews and automated validation wherever practical.

04.5 / Multi-Region Forwarding

How Traffic Traverses the Cloud WAN Core

Consider an application in a production VPC in Region A communicating with another production VPC in Region C.

The source VPC route table must direct the destination prefix toward its Cloud WAN attachment. The regional core network edge then forwards traffic according to the effective routing and segment policy.

01 Source VPC Application initiates traffic
02 VPC Attachment Traffic enters Cloud WAN
03 Core Edge A Segment routing lookup
04 Global Core Inter-region forwarding
05 Core Edge C Destination attachment reached
ILLUSTRATIVE MULTI-REGION TRAFFIC TRACE SOURCE: Region A Production VPC: 10.10.0.0/16 Host: 10.10.1.25 DESTINATION: Region C Production VPC: 10.40.0.0/16 Host: 10.40.1.50 FORWARDING: [1] Source VPC route table 10.40.0.0/16 → Cloud WAN attachment [2] Core Network Edge A Segment: production Destination prefix: 10.40.0.0/16 [3] Cloud WAN Core Inter-region connectivity [4] Core Network Edge C Destination attachment selected [5] Destination VPC Packet delivered to destination RETURN PATH: Reverse routing must also exist. RESULT: PERMITTED when segment policy, routing and security controls align.
04.6 / Hybrid Network Integration

Connecting Branches and On-Premises Networks

Cloud WAN can participate in hybrid network architectures connecting AWS workloads with enterprise data centres and distributed branch environments.

Depending on the chosen attachment architecture, connectivity can involve supported VPN, Connect, Transit Gateway and Direct Connect integration patterns.

Engineers must understand which services exchange routes, how BGP operates across the hybrid boundary, and how on-premises prefixes become reachable within the intended Cloud WAN segments.

HYBRID 01
Site-to-Site VPN

Encrypted connectivity for supported hybrid network attachments, subject to the selected VPN architecture and routing configuration.

HYBRID 02
Direct Connect Integration

Private hybrid connectivity through supported Direct Connect gateway and transit integration designs.

HYBRID 03
Branch Routing

Dynamic routing and controlled prefix advertisement between enterprise networks and AWS transit infrastructure.

04.7 / Production Troubleshooting

Diagnosing Cloud WAN Connectivity Failures

Cloud WAN introduces a managed global routing architecture, but engineers still need to validate attachment states, effective segment membership, deployed policy versions, route visibility and bidirectional connectivity.

✓ Verify the core network policy is deployed successfully.
✓ Confirm regional core network edge availability.
✓ Check attachment state and effective segment association.
✓ Validate destination route visibility within the segment.
✓ Inspect segment-sharing and isolation policies.
✓ Verify source and destination VPC route tables.
✓ Review hybrid BGP advertisements and route filtering.
✓ Trace return routing and security inspection paths.

Production investigation rule: Separate policy intent from effective forwarding state. A policy can define a permitted relationship while a missing attachment, route or VPC configuration still prevents actual communication.

04.8 / Engineering Summary

Cloud WAN as a Policy-Driven Global Transit Architecture

AWS Cloud WAN provides a managed global connectivity model built around regional core network edges, network segments, attachments and declarative policy.

The essential engineering model is: source VPC routing → regional attachment → core network edge → segment policy and route selection → global core forwarding → destination attachment → return path.

This model enables scalable multi-region architecture while retaining the need for explicit routing design, segmentation, security and operational validation.

Next section: Network Segmentation and Security Boundaries — Transit Gateway route table isolation, Cloud WAN segments, shared-services access, restricted routing domains and policy-driven security architecture.

Network Insight · Engineering Guide 20 · Lab 04

AWS Cloud WAN Global Topology Designer

Explore a policy-driven global network connecting AWS Regions through Cloud WAN core network edges. Assign VPC attachments to segments, control segment sharing, inspect route reachability, and simulate multi-Region forwarding failures.

Global Core Network Architecture

Three illustrative AWS Regions connected by a managed Cloud WAN core network. Select a regional edge or attachment to inspect it.

AWS CLOUD WAN · GLOBAL CORE NETWORK CORE EDGE eu-west-1 CORE EDGE eu-central-1 CORE EDGE us-east-1 APPLICATION VPC 10.10.0.0/16 DATABASE VPC 10.20.0.0/16 SHARED SERVICES VPC 10.30.0.0/16
Core edge Active path Failure Core connectivity
3 Regional edges
3 Defined segments
Ready Connectivity state
Select a core network edge or VPC attachment to inspect its role in Cloud WAN.

Cloud WAN Policy Engineering Console

Production ↔ Development sharing
Production ↔ Shared Services sharing
Development ↔ Shared Services sharing
All regional edges available
All VPC attachments available
Configure segment assignments and sharing, then execute a global path trace.

Cloud WAN Segment & Attachment Inventory

Attachment Region Segment State

Segment assignment is modeled as a policy decision. In AWS, attachment policies and tags can determine attachment association with Cloud WAN segments.

Cloud WAN Forwarding Pipeline

1. Identify source attachment and Region
2. Resolve source segment
3. Evaluate segment sharing policy
4. Resolve destination regional edge
5. Forward through Cloud WAN core network
6. Deliver to destination attachment

Global Network Diagnostic Console

[NETWORK INSIGHT · CLOUD WAN LAB 04] Global network initialized. Core network edges: 3 VPC attachments: 3 Segments: production, development, shared Status: READY Awaiting path simulation.

Architecture Interpretation

Global Network: The logical container for Cloud WAN core network resources.
Core Network: The managed network connecting configured regional edges.
Segments: Logical routing domains that support network segmentation.
Attachments: Connect VPCs and supported external connectivity resources to the core.

Production Engineering Notes

AWS Cloud WAN uses a global network, core network, core network policy, regional core network edges, segments, and attachments to provide managed connectivity across AWS Regions and external networks. The core network policy determines segment definitions, edge locations, attachment association behavior, and segment sharing.

Segment sharing controls route visibility between segments according to the configured policy. It should not be interpreted as an automatic firewall rule or universal permission for all traffic. VPC subnet routes, attachment state, security controls, and route availability must also be validated.

This is an educational Cloud WAN topology and policy simulator, not an AWS API-connected network emulator. Regional edge availability, segment sharing, and attachment failures are simplified. The visual core links are logical representations, not physical AWS backbone paths.

Engineering Guide 20 · Editorial Section 05

Network Segmentation and Security Boundaries

Network segmentation is one of the most important design objectives in enterprise AWS transit architectures. Connectivity must be deliberately controlled so that production, development, shared-services and restricted environments communicate only through approved routing and security relationships.

AWS Transit Gateway and AWS Cloud WAN provide different mechanisms for building these boundaries. Transit Gateway uses attachment associations and route tables, while Cloud WAN uses segments and core network policies. Both require careful routing, return-path validation and complementary security controls.

05.1 / Segmentation Architecture

Building Isolated Routing Domains

A segmented enterprise network separates workloads into logical routing domains. Production systems may require access to shared identity, DNS and monitoring services, while development environments should remain isolated from sensitive production resources.

The architectural challenge is to provide approved connectivity without creating unrestricted transitive reachability between environments.

AWS ENTERPRISE TRANSIT — SEGMENTED ROUTING ARCHITECTURE PRODUCTION DOMAIN SHARED SERVICES DOMAIN DEVELOPMENT DOMAIN RESTRICTED DOMAIN Production VPC 10.10.0.0/16 Shared Services 10.20.0.0/16 Development VPC 10.30.0.0/16 Restricted VPC 10.40.0.0/16 Transit Routing Control Plane TGW Route Tables or Cloud WAN Segments × CYAN / GREEN: APPROVED SHARED-SERVICE PATHS · RED: RESTRICTED ROUTING BOUNDARY
Figure 5 — Conceptual segmented AWS transit architecture. Production and development environments can reach approved shared services while restricted networks remain isolated through routing policy.

Engineering principle: Network segmentation should be designed around explicitly permitted communication relationships. A shared transit hub does not require every attached network to communicate with every other network.

05.2 / Transit Gateway Segmentation

Route Table Isolation with AWS Transit Gateway

Transit Gateway route tables act as separate forwarding domains. Each attachment is associated with a Transit Gateway route table, and traffic entering through that attachment is evaluated against the routes in the associated table.

Production and development attachments can be associated with different route tables. Routes to shared services may be installed in both tables, while routes between production and development remain absent.

Transit Gateway Route Table Associated Attachments Permitted Routes Isolation Objective
Production-RT Production VPC attachments Production and Shared Services No direct Development or Restricted routes.
Development-RT Development VPC attachments Development and Shared Services No direct Production or Restricted routes.
Shared-RT Shared Services attachments Approved Production and Development prefixes Controlled return connectivity.
Restricted-RT Restricted workload attachments Restricted networks and approved inspection paths Strictly limited reachability.

This design requires deliberate route propagation and static route configuration. Route visibility must be evaluated in both directions, particularly when multiple environments access a shared service.

ILLUSTRATIVE TGW SEGMENTATION PRODUCTION-RT 10.10.0.0/16 → Production attachment 10.20.0.0/16 → Shared Services attachment DEVELOPMENT-RT 10.30.0.0/16 → Development attachment 10.20.0.0/16 → Shared Services attachment SHARED-RT 10.10.0.0/16 → Production attachment 10.30.0.0/16 → Development attachment RESTRICTED-RT 10.40.0.0/16 → Restricted attachment EXPECTED BEHAVIOUR Production → Shared Services: PERMITTED Development → Shared Services: PERMITTED Production → Development: NO ROUTE Development → Production: NO ROUTE Production → Restricted: NO ROUTE ASSUMPTION: VPC routes and security policy are aligned.

Important security consideration: A shared-services VPC with routes to multiple environments can become a transit point if it contains forwarding appliances or hosts configured to route traffic. Routing isolation should therefore be combined with host, firewall and network security controls.

05.3 / Cloud WAN Segmentation

Global Isolation Through Cloud WAN Segments

AWS Cloud WAN introduces policy-driven segmentation across the global core network. Segments provide logical routing domains that can span multiple configured AWS Regions.

Attachments are associated with segments according to core network policy rules. Separate segments remain isolated unless supported policy configuration explicitly permits route sharing or other approved connectivity.

This model enables an enterprise to apply consistent routing-domain definitions across multiple Regions rather than independently reproducing every regional Transit Gateway route-table relationship.

SEGMENT 01
Production

Contains production VPCs and approved production connectivity across Regions. Access to other segments is explicitly controlled through policy.

SEGMENT 02
Development

Provides a separate routing domain for non-production workloads, reducing unintended connectivity to production environments.

SEGMENT 03
Shared Services

Hosts central services such as DNS, identity and operations platforms. Connectivity is permitted through deliberate segment policy.

ILLUSTRATIVE CLOUD WAN SEGMENT POLICY { "version": "2021.12", "segments": [ { "name": "production", "require-attachment-acceptance": true }, { "name": "development", "require-attachment-acceptance": true }, { "name": "shared-services", "require-attachment-acceptance": true } ], "segment-actions": [ { "action": "share", "mode": "attachment-route", "segment": "shared-services", "share-with": [ "production", "development" ] } ] } INTENT: Share approved Shared Services routes with Production and Development. IMPORTANT: Route sharing is not a firewall. Review reciprocal reachability, attachment policies and route tables. NOTE: Simplified policy fragment for learning. A deployable core network policy requires the complete supported configuration.

Cloud WAN policy changes should be reviewed as network security changes. Segment-sharing configuration can expand reachability across multiple Regions, making unintended route exposure particularly significant.

05.4 / Shared Services Architecture

Controlled Access Without Full Mesh Connectivity

Shared-services networks commonly host central DNS, identity, monitoring, patch management, logging and operational systems.

Production and development workloads may require access to selected services without requiring direct network connectivity to one another.

Engineers should distinguish routing reachability from service-level authorisation. A route to a shared-services VPC does not mean every application should be permitted to access every service in that VPC.

Traffic Relationship Routing Decision Security Requirement
Production → DNS Approved Shared Services route Permit required DNS traffic.
Development → Monitoring Approved Shared Services route Restrict to authorised endpoints and ports.
Production → Development No direct route Maintain isolation and prevent unintended transit.
Restricted → Shared Services Explicitly approved path only Apply enhanced inspection and access controls.
05.5 / Security Enforcement

Routing Isolation Is Not a Complete Security Boundary

Transit Gateway and Cloud WAN provide powerful routing-domain isolation, but network security requires additional enforcement mechanisms.

Security groups control permitted traffic to supported AWS resources. Network ACLs provide subnet-level stateless filtering. Centralised inspection architectures can introduce stateful firewall policy and threat detection.

CONTROL 01
Routing Segmentation

Limit destination reachability through Transit Gateway route tables or Cloud WAN segments.

CONTROL 02
Network Security Policy

Use security groups, network ACLs and firewalls to restrict traffic between authorised workloads and services.

CONTROL 03
Identity and Application Controls

Enforce workload identity, authentication, authorisation and application-level access restrictions independently of network reachability.

Zero Trust design principle: Being connected to the same AWS transit architecture should not establish implicit trust. Each communication relationship should be explicitly justified, constrained and observable.

05.6 / Route Leaks and Segmentation Failures

How Security Boundaries Fail in Production

Routing-domain segmentation can fail when route propagation is enabled unintentionally, attachments are associated with incorrect route tables, or Cloud WAN policy changes expose routes to unintended segments.

Failure Impact Investigation
Incorrect TGW association Attachment uses an unintended routing domain. Inspect attachment-to-route-table association.
Unexpected propagation Additional destination prefixes become visible. Audit propagated routes and propagation settings.
Broad static route Traffic may reach an unintended transit attachment. Inspect longest-prefix matching and route targets.
Cloud WAN segment sharing error Unintended cross-segment route visibility. Compare intended policy with deployed policy and effective routes.
Shared-services transit Forwarding appliances bridge isolated environments. Inspect forwarding, firewall rules and return paths.
Inspection bypass Traffic avoids intended security controls. Trace the complete forwarding and return paths.
ILLUSTRATIVE SEGMENTATION INCIDENT EXPECTED: Production → Shared Services: ALLOW Production → Development: NO ROUTE OBSERVED: Production can reach Development. INVESTIGATION: [1] Identify Production ingress attachment. [2] Inspect associated TGW route table. [3] Find unexpected route: 10.30.0.0/16 → Development attachment [4] Identify source: Unintended route propagation enabled. [5] Review reverse path and security policy. [6] Correct propagation configuration. [7] Revalidate permitted and denied flows. ROOT CAUSE: Unexpected route visibility expanded the Production routing domain. NOTE: The existence of a route alone does not prove an application flow is permitted.
05.7 / Production Validation

Validating Network Segmentation End to End

Segmentation validation should include both positive and negative tests. Engineers must prove that approved communication works and that unapproved communication is blocked.

✓ Verify Transit Gateway route-table associations and propagated routes.
✓ Confirm Cloud WAN segment membership and deployed core network policy.
✓ Validate permitted shared-services connectivity.
✓ Test prohibited cross-environment communication.
✓ Inspect both forward and reverse routing decisions.
✓ Review security groups, NACLs and firewall rules.
✓ Detect unexpected route propagation and overlapping prefixes.
✓ Monitor configuration changes and segmentation drift.

Production validation rule: A segmentation design is not proven simply because permitted traffic succeeds. Negative testing is essential to demonstrate that prohibited paths remain unreachable.

05.8 / Engineering Summary

Segmentation Defines the Security Architecture

AWS Transit Gateway uses route-table association and route visibility to establish regional routing domains. AWS Cloud WAN uses policy-driven segments to establish routing domains across the global core network.

Both approaches support controlled connectivity, but neither should be treated as a replacement for security inspection, workload identity or application-level authorisation.

The essential design objective is: permit only approved routes, enforce appropriate security controls, validate return paths, and continuously detect unintended connectivity.

Next section: Centralised Inspection and AWS Network Firewall — inspection VPCs, Gateway Load Balancer, Transit Gateway appliance mode, Cloud WAN service insertion, symmetric routing and stateful firewall troubleshooting.

Network Insight · Engineering Guide 20 · Lab 05

Network Segmentation Policy Lab

Compare Transit Gateway route-table segmentation with AWS Cloud WAN segment-based connectivity. Define which environments may exchange routes, investigate accidental route leaks, and validate intended isolation between production, development, and shared services.

Interactive Segmentation Architecture

Three security domains connect to a common transit architecture. The forwarding policy determines whether their routes are visible.

ENTERPRISE CLOUD NETWORK · SEGMENTATION PRODUCTION VPC 10.10.0.0/16 Critical workloads DEVELOPMENT VPC 10.20.0.0/16 Engineering workloads TRANSIT GATEWAY Policy Decision SHARED SERVICES VPC 10.30.0.0/16
Production Development Shared services Blocked path
3Security domains
0 Permitted directional paths
Ready Policy state
Select a security domain to inspect its routing and segmentation role.

Segmentation Engineering Console

TGW Route Visibility

Toggle destination routes installed in the ingress attachment's associated route table.

Production → Development route
Development → Production route
Production → Shared route
Shared → Production route
Development → Shared route
Shared → Development route
Blackhole production → development
Configure routing policy and evaluate connectivity between two domains.

Live Directional Connectivity Matrix

Each cell shows whether a modeled route exists from the row's source domain toward the column's destination domain.

A permitted route does not guarantee bidirectional application connectivity. Return routing and security policies must also be validated.

Security Boundary Validation

Expected: Production and Development remain isolated from each other.
Expected: Shared Services can be reached through explicitly approved routes.
Expected: No unapproved route leaks.

Segmentation Diagnostic Console

[NETWORK INSIGHT · SEGMENTATION LAB 05] Mode: Transit Gateway Domains: production, development, shared Status: READY Awaiting policy evaluation.

Engineering Interpretation

TGW segmentation: Different ingress route tables can expose different destination routes.
Cloud WAN segmentation: Core network policy controls segment association and route sharing.
Security boundaries: Routing isolation and packet filtering are separate security mechanisms.

Production Engineering Notes

Transit Gateway segmentation is typically implemented using separate TGW route tables, attachment associations, controlled propagation, and static routes where appropriate. A route-table association determines the forwarding table used by traffic entering through that attachment.

AWS Cloud WAN segments provide logical routing domains. Core network policy defines segments, attachment association, and route sharing. Directional route sharing can create asymmetric route visibility, so engineers should verify both forwarding directions.

Shared-services architectures require careful design. Providing route reachability to a shared-services VPC does not automatically authorize application traffic or permit transitive connectivity between isolated environments.

Educational model: controls represent effective route visibility, not literal AWS API configuration. TGW association and propagation are abstracted, and Cloud WAN policy behavior is simplified. The lab does not simulate every AWS routing precedence rule, service-insertion option, firewall policy, or packet-processing detail.

Engineering Guide 20 · Editorial Section 06

Centralised Inspection and AWS Network Firewall

Centralised inspection introduces security enforcement into AWS transit architectures. Rather than allowing every VPC to communicate directly through the transit network, selected traffic flows can be steered through dedicated inspection infrastructure where firewall policies, threat detection and stateful session controls are applied.

Production inspection designs must coordinate Transit Gateway route tables, inspection VPC routing, AWS Network Firewall, Gateway Load Balancer, appliance mode and, where applicable, Cloud WAN service insertion. The central challenge is maintaining predictable, symmetric forwarding through stateful security devices.

06.1 / Inspection Architecture

Centralised Security Inspection in AWS

A common enterprise design uses a dedicated inspection VPC connected to Transit Gateway. Application VPC traffic is routed through the inspection attachment before reaching its destination.

The inspection VPC may use AWS Network Firewall endpoints or a third-party firewall architecture integrated through Gateway Load Balancer. These are distinct inspection approaches: AWS Network Firewall is a managed firewall service, while Gateway Load Balancer distributes traffic to compatible virtual network appliances.

AWS TRANSIT GATEWAY — CENTRALISED SECURITY INSPECTION APPLICATION VPC Application Workload 10.10.0.0/16 DESTINATION VPC Shared Services 10.20.0.0/16 Transit Gateway Inspection routing CENTRALISED INSPECTION VPC Firewall Endpoint AWS Network Firewall Inspection Policy Stateful rules CYAN: FORWARD INSPECTION PATH · GOLD: RETURN PATH
Figure 6 — Conceptual centralised inspection architecture. The firewall endpoint and policy blocks represent logical inspection functions, not a literal packet path through AWS Network Firewall's internal implementation.

Engineering principle: A centralised firewall is effective only when the intended traffic actually traverses the inspection path. Route tables, attachment behaviour and return-path symmetry must be validated together.

06.2 / AWS Network Firewall

Managed Stateful Inspection

AWS Network Firewall provides managed network traffic filtering using stateless and stateful rule processing. Firewall endpoints are deployed into selected subnets and Availability Zones within a VPC.

Traffic must be routed through the relevant firewall endpoints for inspection. Firewall policies can evaluate traffic based on supported network and application attributes, depending on the rule configuration.

FIREWALL 01
Stateless Rules

Evaluate packets using stateless matching criteria and configured actions before applicable stateful processing.

FIREWALL 02
Stateful Rules

Track supported network sessions and enforce stateful filtering, including rules based on protocol and application traffic characteristics.

FIREWALL 03
Logging and Visibility

Generate supported alert and flow logs for security analysis, incident investigation and operational monitoring.

Stateful inspection depends on session consistency. If forward traffic passes through one firewall endpoint while return traffic unexpectedly traverses another stateful inspection context, the return packets may be dropped.

06.3 / Gateway Load Balancer

Third-Party Firewall Appliance Integration

Gateway Load Balancer provides an architecture for deploying and scaling compatible virtual network appliances. It can distribute traffic across registered appliance targets and support transparent insertion into network forwarding paths.

In contrast to AWS Network Firewall, Gateway Load Balancer is not itself a managed firewall rule engine. Security inspection is performed by the compatible appliances behind the load balancer.

Capability AWS Network Firewall Gateway Load Balancer
Primary role Managed network firewall inspection. Traffic distribution and insertion for compatible appliances.
Security rules Configured through Network Firewall policies. Implemented by the third-party appliance.
Scaling model AWS-managed firewall service architecture. Registered appliance targets with load-balancing integration.
Routing requirement Traffic must traverse firewall endpoints. Traffic must traverse the appropriate GWLB endpoint and appliance path.
06.4 / Transit Gateway Appliance Mode

Preserving Stateful Traffic Symmetry

Transit Gateway appliance mode is an attachment option used in supported appliance VPC architectures. It helps maintain consistent Availability Zone selection for both directions of a flow traversing a stateful appliance attachment.

Without appropriate appliance-mode and routing design, forward and return packets may traverse different inspection paths, particularly in multi-AZ architectures.

ILLUSTRATIVE STATEFUL INSPECTION FLOW FORWARD: Application VPC ↓ Transit Gateway ↓ Inspection VPC — AZ A ↓ Firewall Endpoint — AZ A ↓ Transit Gateway ↓ Destination VPC RETURN: Destination VPC ↓ Transit Gateway ↓ Inspection VPC — AZ A ↓ Same logical inspection context ↓ Transit Gateway ↓ Application VPC EXPECTED: Stateful session continuity. IMPORTANT: Appliance mode assists with consistent AZ selection on the appliance attachment. It does not replace correct routing, firewall policy or endpoint design.

Critical distinction: Appliance mode helps preserve flow symmetry at the Transit Gateway attachment level. It does not automatically guarantee that every internal firewall path or appliance session is symmetric.

06.5 / Cloud WAN Service Insertion

Policy-Driven Inspection in Global Networks

AWS Cloud WAN supports service insertion capabilities for supported network architectures. These allow engineers to define traffic steering through approved inspection or network function attachments using core network policy.

This approach can reduce the need to maintain independent inspection routing configurations across multiple regional transit networks. However, the effective behaviour depends on supported attachment types, policy configuration, regional deployment and routing.

DESIGN 01
Inspection Segment

A dedicated routing domain associated with security inspection infrastructure.

DESIGN 02
Service Insertion Policy

Supported policy configuration determines which traffic relationships require inspection.

DESIGN 03
Regional Inspection

Inspection placement and attachment design influence latency, resilience and forwarding symmetry.

Engineers should verify supported service insertion patterns and regional availability before adopting a design. Not every firewall deployment or attachment combination has identical service insertion behaviour.

06.6 / Inspection Packet Pipeline

From Workload to Firewall to Destination

01 Source VPC Traffic enters transit
02 Transit Routing Inspection path selected
03 Firewall Endpoint Security policy evaluated
04 Destination Routing Approved traffic forwarded
05 Return Inspection Session symmetry validated

A production inspection architecture must ensure that both directions of a stateful session traverse the intended security controls. This includes verifying Transit Gateway or Cloud WAN routing, inspection VPC subnet routes, firewall endpoint placement and destination return routing.

06.7 / Failure Investigation

Common Centralised Inspection Failures

Failure Symptom Investigation
Inspection bypass Traffic reaches destination without expected inspection. Review transit routes and service insertion policy.
Asymmetric forwarding Stateful firewall drops return packets. Inspect appliance mode, AZ routing and firewall session paths.
Missing firewall endpoint route Traffic fails inside inspection VPC. Review inspection subnet route tables and endpoint targets.
Incorrect firewall policy Legitimate application sessions are denied. Inspect firewall rule evaluation and alert logs.
GWLB appliance health failure Reduced inspection capacity or interrupted flows. Inspect target health and appliance availability.
Cloud WAN insertion policy error Unexpected inspection path or missing connectivity. Validate deployed core network policy and effective routing.
PRODUCTION INSPECTION INCIDENT SYMPTOM: Application establishes outbound traffic, but return sessions fail intermittently. INVESTIGATION: [1] Confirm source VPC route to transit. [2] Identify transit route table or Cloud WAN segment forwarding policy. [3] Trace forward inspection attachment. [4] Identify firewall endpoint / appliance. [5] Trace destination return route. [6] Compare forward and return AZ paths. [7] Review firewall stateful session logs. [8] Validate appliance mode and inspection VPC routing. POSSIBLE ROOT CAUSE: Forward and return traffic traverse different stateful inspection contexts. REMEDIATION: Correct routing and inspection symmetry, then revalidate the complete session.
06.8 / Production Validation

Validating a Secure Inspection Architecture

✓ Confirm all required traffic traverses the inspection path.
✓ Validate forward and reverse Transit Gateway routing.
✓ Review appliance mode for supported inspection attachments.
✓ Verify firewall endpoint Availability Zone coverage.
✓ Confirm Network Firewall policy and logging configuration.
✓ Validate GWLB target health where appliances are used.
✓ Inspect Cloud WAN service insertion policy where applicable.
✓ Test both permitted and denied traffic through the inspection path.

Production validation rule: A successful connection does not prove that the intended firewall inspected the traffic. Validate the actual forwarding path and correlate it with inspection logs.

06.9 / Engineering Summary

Secure Transit Requires Controlled Inspection Paths

Centralised inspection introduces a security enforcement layer into AWS transit networking. Transit Gateway route tables, appliance mode, inspection VPC routing and firewall endpoint placement must operate together.

AWS Cloud WAN can extend this model through supported policy-driven service insertion architectures. Regardless of the implementation, engineers must preserve traffic steering, firewall policy enforcement, stateful symmetry and operational visibility.

Next section: Hybrid Networking with AWS Direct Connect and Site-to-Site VPN — Direct Connect gateways, Transit Gateway integration, BGP route exchange, branch connectivity, hybrid failover and production troubleshooting.

Network Insight · Engineering Guide 20 · Lab 06

Stateful Firewall Traffic Flow Simulator

Investigate centralised AWS traffic inspection through a Transit Gateway-connected inspection VPC. Explore AWS Network Firewall, Gateway Load Balancer appliance paths, stateful session tracking, appliance mode, and symmetric return routing.

Dual-AZ Centralised Inspection Architecture

The topology illustrates source and destination VPCs, a regional Transit Gateway, and an inspection VPC containing two inspection zones.

AWS REGION · CENTRAL INSPECTION APPLICATION VPC 10.10.0.0/16 Source workload DATABASE VPC 10.20.0.0/16 Destination workload TRANSIT GATEWAY Route Tables INSPECTION VPC · 10.40.0.0/16 INSPECTION ZONE A Firewall Endpoint A Stateful session engine INSPECTION ZONE B Firewall Endpoint B Independent session state
Transit Gateway Inspection path Failure
NEW TCP session state
AZ A Selected inspection zone
Ready Inspection status
Select a network component to inspect its role in centralised traffic inspection.

Stateful Inspection Engineering Console

Inspection Routing Controls
TGW appliance mode enabled
Forward route through inspection
Return route through inspection
Firewall policy permits selected port
Inspection endpoints healthy
Return path uses same inspection zone
Configure the inspection path and simulate a TCP session through the firewall.

TCP Session State Machine

1. TCP SYN enters Transit Gateway
2. TGW routes packet into inspection VPC
3. Firewall evaluates policy and creates state
4. SYN reaches destination workload
5. SYN-ACK returns through inspection
6. Session becomes established

Inspection Routing and Session Table

Flow Inspection zone Decision

Stateful inspection requires compatible forward and return paths. Appliance mode supports flow affinity for supported Transit Gateway appliance architectures, but correct route tables remain essential.

Firewall Diagnostic Console

[NETWORK INSIGHT · FIREWALL LAB 06] Inspection architecture initialized. Transit Gateway: tgw-core Inspection VPC: 10.40.0.0/16 Firewall zones: A, B Session state: NEW Awaiting TCP session simulation.

Engineering Interpretation

Central inspection: Traffic is steered into a dedicated inspection VPC before reaching its destination.
Stateful firewall: Return traffic must reach a compatible stateful inspection context.
Appliance mode: Helps maintain consistent AZ selection for a flow through a TGW VPC attachment.

Production Engineering Notes

AWS Network Firewall is a managed, stateful network security service deployed using firewall endpoints in selected Availability Zones. A centralised inspection design requires routing that steers relevant traffic through these endpoints.

Gateway Load Balancer is a separate service-insertion architecture for deploying compatible third-party virtual appliances. It distributes traffic to appliance targets and supports flow stickiness. It should not be treated as the same service as AWS Network Firewall.

Transit Gateway appliance mode helps maintain flow affinity to a selected Availability Zone through an appliance VPC attachment. It does not automatically correct missing VPC routes, incorrect TGW route tables, or all asymmetric-routing problems.

Educational simulation only. The model represents a simplified stateful inspection workflow. It does not emulate actual AWS Network Firewall endpoint selection, Suricata rule evaluation, GWLB GENEVE encapsulation, appliance target selection, TCP retransmissions, or AWS control-plane behavior.

Engineering Guide 20 · Editorial Section 07

Hybrid Networking: AWS Direct Connect and Site-to-Site VPN

Hybrid AWS networking connects enterprise data centres, branch offices and private infrastructure to cloud workloads through dedicated connectivity, encrypted VPN tunnels and dynamic routing.

AWS Direct Connect and Site-to-Site VPN serve different connectivity requirements. Direct Connect provides dedicated connectivity into AWS, while Site-to-Site VPN provides encrypted IPsec tunnels over supported transport networks. Both can integrate with AWS transit architectures to provide access to multiple VPCs and AWS Regions.

Production hybrid design requires understanding Direct Connect gateways, transit virtual interfaces, BGP routing, Transit Gateway, Cloud WAN integration and failover behaviour.

07.1 / Hybrid Connectivity Architecture

Connecting Enterprise Networks to AWS Transit

A common enterprise hybrid architecture uses a customer router connected to AWS Direct Connect. A transit virtual interface can connect to a Direct Connect gateway, which in turn can associate with supported Transit Gateway resources.

Site-to-Site VPN can provide an alternative connectivity path. VPN routing can use BGP where supported and configured, allowing route exchange between the enterprise network and AWS.

ENTERPRISE HYBRID NETWORK — DIRECT CONNECT + VPN + AWS TRANSIT ON-PREMISES NETWORK Enterprise Router BGP / WAN Edge Data Centre 172.16.0.0/16 Direct Connect Gateway Transit VIF / DXGW Site-to-Site VPN IPsec / BGP AWS Transit Gateway Regional routing hub AWS APPLICATION VPCS Production 10.10.0.0/16 Shared Services 10.20.0.0/16 CYAN: DIRECT CONNECT PATH · PURPLE: VPN ALTERNATIVE PATH
Figure 7 — Conceptual AWS hybrid architecture. Direct Connect and VPN are represented as alternative paths. Actual failover depends on BGP policy, AWS route selection, attachment configuration and path health.

Engineering principle: Direct Connect and VPN provide transport connectivity, but end-to-end hybrid reachability still depends on BGP advertisements, Transit Gateway or Cloud WAN routing, VPC subnet routes and return-path configuration.

07.2 / AWS Direct Connect

Dedicated Private Connectivity to AWS

AWS Direct Connect provides dedicated network connectivity between an enterprise environment and AWS. It can support private access patterns without relying exclusively on internet-based VPN transport.

For Transit Gateway integration, a common design uses a transit virtual interface associated with a Direct Connect gateway. The Direct Connect gateway can associate with supported Transit Gateway resources, subject to AWS routing and association rules.

DX COMPONENT 01
Direct Connect Connection

Dedicated physical connectivity between the customer network and AWS Direct Connect infrastructure.

DX COMPONENT 02
Transit Virtual Interface

Logical interface used for supported Transit Gateway connectivity through a Direct Connect gateway.

DX COMPONENT 03
Direct Connect Gateway

Provides a connectivity abstraction for supported private and transit gateway association architectures.

Direct Connect is not inherently encrypted end to end. Organisations requiring encryption should evaluate supported MACsec, VPN-over-Direct-Connect or application-layer encryption designs as appropriate.

07.3 / AWS Site-to-Site VPN

Encrypted Hybrid Connectivity

AWS Site-to-Site VPN establishes encrypted IPsec connectivity between a customer gateway and a supported AWS VPN termination architecture.

Transit Gateway VPN attachments can provide hybrid access to multiple attached VPCs when the routing configuration permits it.

Dynamic VPN routing commonly uses BGP to advertise reachable network prefixes and respond to route changes. Static routing is also supported in appropriate VPN configurations.

Characteristic Direct Connect Site-to-Site VPN
Transport Dedicated connectivity into AWS. IPsec tunnels over supported transport.
Encryption Not inherently encrypted end to end. IPsec encryption.
Routing BGP over supported virtual interfaces. BGP or supported static routing.
Typical role Primary private hybrid connectivity. Primary or alternative hybrid connectivity.
Resilience Requires deliberate redundant connection design. Multiple tunnels and suitable routing design.
07.4 / Hybrid BGP Routing

Route Advertisements and Prefix Control

BGP enables route exchange between supported AWS hybrid networking services and enterprise routers. The on-premises router advertises reachable enterprise prefixes, while AWS advertises permitted cloud network prefixes according to the chosen service architecture.

Engineers must distinguish BGP route learning from Transit Gateway route-table association and propagation. These are different routing mechanisms that interact within the overall hybrid forwarding architecture.

ILLUSTRATIVE HYBRID ROUTE EXCHANGE ON-PREMISES: Enterprise ASN: 65010 Advertised prefix: 172.16.0.0/16 AWS: Transit Gateway ASN: 64512 Cloud VPC prefixes: 10.10.0.0/16 10.20.0.0/16 BGP CONTROL PLANE: Enterprise router advertises 172.16.0.0/16 through supported DX/VPN routing sessions. AWS TRANSIT ROUTING: Hybrid routes become available according to attachment and propagation configuration. VPC FORWARDING: Subnet route tables must direct on-premises destinations toward the appropriate transit target. RETURN PATH: Enterprise routers must learn and select routes toward permitted AWS VPC prefixes. NOTE: Actual route advertisements, permitted prefixes and selection depend on AWS service configuration.

BGP engineering rule: An established BGP session does not guarantee end-to-end connectivity. Verify advertised prefixes, accepted routes, Transit Gateway route tables, VPC routes and the reverse path.

07.5 / Cloud WAN Hybrid Integration

Extending Hybrid Connectivity Across Regions

Cloud WAN can provide a global transit architecture for hybrid-connected environments. Supported VPN, Connect and Transit Gateway integration patterns can connect external networks into the core network.

Direct Connect integration requires a supported connectivity architecture rather than assuming every Direct Connect gateway attaches directly to Cloud WAN in the same way as a Transit Gateway.

Cloud WAN segments can then define which regional workloads may reach the hybrid network. This provides a policy-driven model for controlling enterprise network connectivity across multiple AWS Regions.

DESIGN 01
Regional Hybrid Entry

External connectivity enters the AWS architecture through a supported regional attachment or transit integration.

DESIGN 02
Global Core Forwarding

Cloud WAN provides managed multi-region transit according to effective routing and segment policies.

DESIGN 03
Segment-Based Access

Production, development and restricted segments can receive different hybrid routing permissions.

07.6 / Hybrid Resilience

Direct Connect and VPN Failover Design

Hybrid resilience requires more than configuring a second connectivity path. Engineers must understand route selection, tunnel state, BGP advertisements, failover timing and return-path behaviour.

A common architecture uses Direct Connect as the preferred path and Site-to-Site VPN as an alternative path. However, the exact preference is determined by supported AWS route priorities and the configured routing policy.

01 Primary Path DX connectivity available
02 Failure Event DX path becomes unavailable
03 Route Withdrawal BGP and service state update
04 Alternative Path VPN route becomes selected
05 Validation Bidirectional traffic tested
ILLUSTRATIVE HYBRID FAILOVER NORMAL STATE: Direct Connect: AVAILABLE VPN: AVAILABLE Preferred path: Direct Connect EVENT: Direct Connect connectivity fails. CONTROL PLANE: Primary routes become unavailable. Alternative route is evaluated. EXPECTED RESULT: VPN path carries permitted traffic when routing policy and path health support failover. VALIDATION: Confirm AWS → on-premises path. Confirm on-premises → AWS path. Verify encryption and firewall policy. Measure application recovery time. IMPORTANT: Failover is not instantaneous. Route convergence and session recovery depend on the design.
07.7 / Production Troubleshooting

Diagnosing Hybrid AWS Connectivity Failures

Failure Symptom Investigation
Direct Connect physical failure Primary hybrid path unavailable. Inspect connection state and physical connectivity.
BGP session failure Hybrid routes missing or withdrawn. Inspect BGP neighbour state and advertisements.
Missing TGW propagation On-premises prefix absent from required route table. Inspect attachment propagation configuration.
Missing VPC route Cloud workload cannot reach enterprise network. Inspect source subnet route table.
VPN tunnel failure Alternative connectivity unavailable. Inspect tunnel status, IKE/IPsec and routing.
Asymmetric hybrid routing Sessions fail despite partial connectivity. Compare AWS and enterprise return paths.
✓ Validate Direct Connect connection and virtual interface state.
✓ Confirm BGP neighbour establishment and route advertisements.
✓ Inspect Direct Connect gateway and Transit Gateway association configuration.
✓ Verify VPN tunnel state and backup route availability.
✓ Check Transit Gateway or Cloud WAN route visibility.
✓ Confirm VPC subnet routes to hybrid prefixes.
✓ Trace reverse routing from enterprise networks.
✓ Test failure and recovery under controlled conditions.

Production investigation rule: Hybrid connectivity failures should be traced across the complete path: physical or VPN transport, BGP control plane, AWS transit routing, VPC routing, security policy and enterprise return routing.

07.8 / Engineering Summary

Hybrid AWS Networking Is a Multi-Layer Routing System

AWS Direct Connect and Site-to-Site VPN provide complementary methods for connecting enterprise networks to AWS.

Direct Connect gateways, Transit Gateway routing, Cloud WAN integration and BGP route exchange determine how hybrid networks become reachable across AWS environments.

The essential design model is: transport connectivity → BGP route exchange → transit route selection → VPC routing → security enforcement → return-path validation.

Next section: Multi-Region Networking and Resilience — Transit Gateway peering, Cloud WAN regional core edges, global routing, failure domains, route convergence and production failover investigation.

Network Insight · Engineering Guide 20 · Lab 07

Hybrid AWS Connectivity Simulator

Explore hybrid connectivity from an enterprise data centre to AWS through Direct Connect, Direct Connect gateways, Transit Gateway, and Site-to-Site VPN. Investigate BGP route advertisements, preferred paths, failover, and missing-route incidents.

Enterprise Hybrid Connectivity Architecture

The logical topology includes an on-premises enterprise router, a primary Direct Connect path, a backup Site-to-Site VPN path, and AWS Transit Gateway connectivity.

HYBRID AWS NETWORK · LOGICAL TOPOLOGY ENTERPRISE DATA CENTRE 172.16.0.0/16 Customer Gateway · BGP DIRECT CONNECT GATEWAY Transit VIF · BGP SITE-TO-SITE VPN IPsec · BGP Backup Connectivity AWS TRANSIT GATEWAY Hybrid Route Selection AWS APPLICATION VPC 10.10.0.0/16 eu-west-1
Direct Connect VPN backup Selected path Unavailable
DX Selected hybrid path
2 Eligible path advertisements
Ready Connectivity status
Select a network component to inspect its hybrid connectivity role.

Hybrid Routing Engineering Console

Hybrid Control Plane
Direct Connect connection operational
Direct Connect BGP route advertised
VPN tunnel operational
VPN BGP route advertised
Return route installed
Configure hybrid connectivity and execute the routing simulation.

Hybrid BGP Route Table

Prefix Path BGP state Selection

The preference selector models an intended enterprise routing policy. Real path selection depends on AWS routing behavior, prefix specificity, route origin, BGP attributes, and supported service-specific route priorities.

Hybrid Packet Forwarding Pipeline

1. Source routing lookup
2. Evaluate BGP route advertisements
3. Select Direct Connect or VPN path
4. Enter AWS Transit Gateway
5. Forward toward destination prefix
6. Validate return-path connectivity

Hybrid Network Diagnostic Console

[NETWORK INSIGHT · HYBRID LAB 07] Enterprise prefix: 172.16.0.0/16 AWS prefix: 10.10.0.0/16 Primary path: Direct Connect Backup path: Site-to-Site VPN BGP state: READY Awaiting traffic simulation.

Engineering Interpretation

Direct Connect: Dedicated connectivity to AWS through a Direct Connect location.
Direct Connect gateway: Connects supported Direct Connect virtual interfaces with associated AWS gateway resources.
Site-to-Site VPN: Provides IPsec tunnel connectivity that can support dynamic BGP routing.
Hybrid resilience: Requires route availability and correct forwarding in both directions.

Production Engineering Notes

AWS Direct Connect provides dedicated network connectivity to AWS. A transit virtual interface connects to a Direct Connect gateway, which can be associated with supported Transit Gateway resources. Direct Connect gateway associations and allowed prefixes must be configured correctly for the intended architecture.

AWS Site-to-Site VPN commonly provides two IPsec tunnels for resilience. Dynamic routing can use BGP to exchange prefixes between a customer gateway and the AWS VPN endpoint.

Failover depends on route availability, service-specific route priorities, routing policy, tunnel health, and convergence behavior. A VPN backup does not guarantee seamless failover without correct routing and operational validation.

Educational simulation only. The preference selector abstracts AWS route priority and BGP policy. This lab does not implement actual AWS Direct Connect BGP decision rules, tunnel negotiation, DX gateway allowed-prefix processing, or control-plane convergence timers.

Engineering Guide 20 · Editorial Section 08

Multi-Region Networking and Resilience

Multi-region AWS networking extends enterprise connectivity beyond a single regional transit architecture. Organisations use multiple AWS Regions to improve availability, distribute applications geographically and support disaster recovery requirements.

AWS Transit Gateway and AWS Cloud WAN provide different approaches to multi-region connectivity. Transit Gateway supports inter-region peering, while Cloud WAN provides a managed global core network with regional core network edges and policy-driven routing.

Resilience depends on more than connecting Regions. Engineers must consider failure domains, route propagation, traffic steering, application recovery, inspection placement and return-path consistency.

08.1 / Multi-Region Architecture

Connecting Regional AWS Networks

A multi-region transit design typically includes independent VPCs and regional networking resources. Transit Gateway peering can connect regional transit gateways, while Cloud WAN can provide managed inter-region connectivity through its core network edges.

The appropriate architecture depends on operational scale, routing complexity, segmentation requirements, supported services and resilience objectives.

AWS MULTI-REGION NETWORKING — TRANSIT GATEWAY / CLOUD WAN REGION A — PRIMARY Regional Transit Hub Production VPC 10.10.0.0/16 REGION B — SECONDARY Regional Transit Hub Recovery VPC 10.20.0.0/16 REGION C — GLOBAL SERVICES Regional Transit Hub Shared Services 10.30.0.0/16 CYAN: PRIMARY INTER-REGION PATH · GREEN: ALTERNATIVE REGIONAL CONNECTIVITY
Figure 8 — Conceptual multi-region AWS transit architecture. The regional hubs represent Transit Gateways or Cloud WAN core network edges depending on the selected design. The illustration does not imply that Transit Gateway peering automatically provides transitive routing through Region B.

Engineering principle: Multi-region connectivity and multi-region resilience are different capabilities. Connectivity enables routing between Regions, while resilience requires redundant paths, suitable routing policies and recoverable applications.

08.2 / Transit Gateway Peering

Inter-Region Connectivity with Transit Gateway

Transit Gateway peering connects two Transit Gateways in supported peering architectures, including inter-region connectivity.

Peering attachments provide a forwarding relationship between the two transit gateways, but route propagation across Transit Gateway peering attachments is not automatic.

Engineers must configure appropriate static routes in the relevant Transit Gateway route tables and ensure that VPC subnet routes support the intended forwarding paths.

ILLUSTRATIVE TGW INTER-REGION ROUTING REGION A: TGW-A Application VPC: 10.10.0.0/16 REGION B: TGW-B Recovery VPC: 10.20.0.0/16 PEERING: TGW-A ↔ TGW-B TGW-A ROUTE TABLE: 10.20.0.0/16 → TGW Peering Attachment TGW-B ROUTE TABLE: 10.10.0.0/16 → TGW Peering Attachment APPLICATION VPC ROUTE: 10.20.0.0/16 → TGW-A RECOVERY VPC ROUTE: 10.10.0.0/16 → TGW-B IMPORTANT: TGW peering routes are configured explicitly; they are not automatically propagated. RESULT: Bidirectional connectivity when routing and security permit.

Important routing limitation: Transit Gateway peering is not a general-purpose transitive routing mechanism. Engineers should not assume that a chain of peered Transit Gateways provides automatic connectivity between non-adjacent Regions.

08.3 / Cloud WAN Global Core

Managed Multi-Region Connectivity

AWS Cloud WAN provides a managed global core network with regional core network edges. The core network policy defines regional locations, network segments, attachment relationships and supported routing controls.

This architecture can simplify the management of global connectivity compared with independently maintaining multiple Transit Gateway peering relationships.

However, Cloud WAN does not eliminate the need for routing validation. Effective segment membership, attachment state, VPC route tables and security policy still determine end-to-end reachability.

Characteristic Transit Gateway Peering AWS Cloud WAN
Architecture Regional Transit Gateways connected through peering. Managed global core network with regional edges.
Routing management Regional route-table configuration and explicit peering routes. Core network policy and segment-based routing.
Segmentation Transit Gateway route tables. Global core network segments.
Operational model Regional resource and peering management. Centralised global policy management.
Resilience Requires explicit architecture and failure planning. Managed core infrastructure, with application and attachment resilience still required.
08.4 / Failure Domain Design

Designing for Regional and Network Failures

Multi-region resilience begins by identifying failure domains. These can include individual Availability Zones, VPC attachments, transit resources, hybrid connectivity paths and entire AWS Regions.

A regional application failure does not necessarily mean the global network has failed. Similarly, a healthy global transit network does not guarantee that applications remain available.

FAILURE DOMAIN 01
Availability Zone

Attachment subnet placement, firewall endpoint coverage and application distribution affect AZ-level resilience.

FAILURE DOMAIN 02
Regional Transit

Regional Transit Gateway resources and Cloud WAN core network edges require suitable architectural recovery planning.

FAILURE DOMAIN 03
Application Region

Recovery requires suitable application deployment, data replication, endpoint selection and traffic redirection.

Resilience design rule: Redundant network connectivity cannot compensate for an application that has no recoverable deployment in the secondary Region.

08.5 / Regional Failover Engineering

From Primary Region to Recovery Region

Regional failover requires coordination between network routing and application recovery mechanisms.

Engineers must determine whether recovery involves switching application endpoints, changing routing advertisements, updating service discovery or redirecting users to a secondary deployment.

01 Normal Operation Primary Region serves traffic
02 Failure Detection Regional health degrades
03 Recovery Decision Failover policy activated
04 Traffic Redirection Recovery endpoints selected
05 Validation Application and network tested
ILLUSTRATIVE MULTI-REGION RECOVERY PRIMARY REGION: eu-west-1 Application VPC: 10.10.0.0/16 RECOVERY REGION: eu-central-1 Recovery VPC: 10.20.0.0/16 NORMAL STATE: Primary application endpoints active. Recovery infrastructure available. EVENT: Primary application Region unavailable. RECOVERY WORKFLOW: Detect application failure. Confirm recovery environment health. Activate approved failover mechanism. Redirect application traffic. Validate hybrid and VPC routing. Verify security and inspection paths. Confirm application recovery. IMPORTANT: Transit Gateway and Cloud WAN do not automatically perform application disaster recovery. Recovery time depends on the complete application architecture.
08.6 / Route Convergence and Recovery

Understanding Control-Plane and Data-Plane Recovery

Network convergence is the process through which routing systems adapt to topology or connectivity changes.

In AWS transit architectures, convergence behaviour depends on the affected service, attachment type, route configuration and supported routing mechanisms.

BGP-based hybrid connections may withdraw or update routes following failures. Static Transit Gateway peering routes do not automatically behave like dynamic BGP advertisements.

Cloud WAN manages parts of the global core network control plane, but engineers must still validate application traffic recovery and effective routing state.

Recovery Layer Primary Concern Validation Method
Physical / Transport Connectivity path availability. Connection, tunnel and interface health.
Control Plane Route withdrawal and selection. BGP state, route tables and effective policy.
Data Plane Packet forwarding and return paths. Traffic tests and flow telemetry.
Security Firewall and inspection continuity. Policy validation and firewall logs.
Application Service availability and recovery. Health checks and application transactions.
08.7 / Production Troubleshooting

Investigating Multi-Region Network Failures

✓ Confirm regional Transit Gateway or Cloud WAN core edge state.
✓ Inspect inter-region peering attachment status.
✓ Validate static peering routes where applicable.
✓ Review Cloud WAN segment and attachment policy.
✓ Verify source and destination VPC subnet routes.
✓ Inspect hybrid BGP route advertisements and withdrawals.
✓ Trace firewall and inspection symmetry.
✓ Validate application failover and recovery.
MULTI-REGION INCIDENT INVESTIGATION SYMPTOM: Region A workloads cannot reach services in Region B. STEP 1: Verify source VPC subnet route. STEP 2: Identify regional transit attachment. STEP 3: Inspect TGW peering route or Cloud WAN segment route. STEP 4: Confirm destination attachment state. STEP 5: Inspect destination VPC routing. STEP 6: Trace reverse forwarding path. STEP 7: Review security and inspection policy. STEP 8: Validate application endpoint health. POSSIBLE ROOT CAUSES: Missing static peering route. Incorrect segment association. Failed attachment. Missing VPC return route. Security policy denial. Application service unavailable.

Production investigation rule: Separate network connectivity failure from application availability failure. A reachable secondary Region does not prove that the application has successfully failed over.

08.8 / Engineering Summary

Multi-Region Resilience Requires Coordinated Design

AWS Transit Gateway peering enables explicit connectivity between regional transit networks, while Cloud WAN provides a managed global core network architecture.

Both approaches require deliberate routing, segmentation, attachment placement, inspection and recovery design.

The essential resilience model is: regional connectivity → route availability → security continuity → application recovery → operational validation.

Next section: Production Troubleshooting and Architecture Validation — missing routes, asymmetric forwarding, attachment failures, Cloud WAN policy errors, inspection bypass, hybrid connectivity incidents and systematic AWS network fault isolation.

Network Insight · Engineering Guide 20 · Lab 08

Multi-Region Failure Investigation Lab

Compare AWS Transit Gateway inter-Region peering with AWS Cloud WAN core network connectivity. Investigate regional outages, missing routes, segmentation failures, and recovery through an alternate application deployment.

Global Multi-Region Network Architecture

Three AWS Regions contain application, database, and disaster-recovery environments connected through the selected transit architecture.

AWS MULTI-REGION NETWORK · RESILIENCE REGIONAL EDGE eu-west-1 REGIONAL EDGE eu-central-1 REGIONAL EDGE us-east-1 APPLICATION VPC 10.10.0.0/16 DATABASE VPC 10.20.0.0/16 DR APPLICATION VPC 10.30.0.0/16
Regional transit edge Active route Failure
3 Regional edges
Primary Application target
Ready Connectivity status
Select a Region or VPC to inspect its role in the multi-Region design.

Multi-Region Resilience Console

Regional Routing Controls
Ireland regional edge operational
Frankfurt regional edge operational
Virginia regional edge operational
Inter-Region connectivity available
Destination route installed
Segment policy permits connectivity
DR failover enabled
Configure the regional architecture and execute a multi-Region route investigation.

Multi-Region Route and Edge Inventory

Region Prefix Edge Route

Transit Gateway inter-Region peering requires appropriate static routing toward peering attachments. Cloud WAN uses core network policy and segment route propagation rather than identical TGW peering route-table configuration.

Regional Failure Investigation Pipeline

1. Validate source Region
2. Resolve destination route
3. Validate inter-Region connectivity
4. Evaluate segment policy
5. Validate destination regional edge
6. Deliver to destination workload

Multi-Region Diagnostic Console

[NETWORK INSIGHT · MULTI-REGION LAB 08] Source Region: eu-west-1 Primary Region: eu-central-1 DR Region: us-east-1 Transit architecture: TGW peering Status: READY Awaiting failure investigation.

Engineering Interpretation

TGW peering: Connects Transit Gateways across AWS Regions using peering attachments.
Cloud WAN: Uses managed regional core network edges and segment policies.
Disaster recovery: Requires a reachable alternate workload and deliberate failover.

Production Engineering Notes

AWS Transit Gateway supports inter-Region peering between Transit Gateways. Routing through a peering attachment requires appropriate route-table configuration; TGW peering does not automatically provide transitive routing across an arbitrary mesh of Transit Gateways.

AWS Cloud WAN provides a managed global core network with regional edges and segment-based routing. Its policy-driven model can simplify connectivity management across multiple Regions, but route propagation and segment boundaries still require careful design.

Multi-Region disaster recovery requires more than network connectivity. Application readiness, data replication, DNS or traffic steering, dependency recovery, security policy, and return routing must be validated before a DR environment can serve production traffic.

Educational simulation only. Regional availability and path selection are simplified. The model does not represent actual AWS backbone paths, TGW route precedence, Cloud WAN route evaluation, service health, replication state, or real convergence timing.

Engineering Guide 20 · Editorial Section 09

Production Troubleshooting and Architecture Validation

Production AWS transit networking requires engineers to understand not only how connectivity is designed, but how to investigate failures across multiple routing domains, attachments, Regions, security controls and hybrid connections.

A successful troubleshooting methodology separates control-plane state, data-plane forwarding, security enforcement and application availability. Each layer must be validated independently before the complete end-to-end path can be considered healthy.

This final section brings together Transit Gateway routing, Cloud WAN segments, VPC attachments, AWS Network Firewall, Direct Connect, VPN and multi-region resilience into a single operational investigation framework.

09.1 / Incident Investigation Architecture

Following the Complete AWS Packet Path

Effective network troubleshooting begins with a clearly defined source, destination and expected application behaviour. Engineers then trace the actual forwarding path, validating every routing and security decision along the way.

In AWS transit architectures, the path can cross VPC route tables, Transit Gateway attachments, Cloud WAN core network edges, inspection VPCs and hybrid routing boundaries.

AWS NETWORK INCIDENT — END-TO-END FORWARDING INVESTIGATION SOURCE VPC Application 10.10.1.25 Transit Gateway Inspection Firewall Stateful policy DESTINATION NETWORK Destination 10.20.1.40 Cloud WAN Core Network Segment / regional edge Hybrid Network DX / VPN / BGP On-premises connectivity CYAN: FORWARD PATH · GOLD: RETURN PATH
Figure 9 — Conceptual AWS transit incident investigation architecture. The Cloud WAN and hybrid components represent alternative or additional connectivity domains rather than mandatory sequential hops in every flow.

Engineering principle: Never assume that an attachment in an available state, an established BGP session or a deployed Cloud WAN policy proves successful end-to-end forwarding. Each layer requires separate validation.

09.2 / Structured Troubleshooting

The Five-Layer AWS Network Investigation Model

01 Endpoint Source and destination state
02 VPC Routing Subnet routes and attachments
03 Transit Routing TGW tables / Cloud WAN segments
04 Security Firewall and inspection controls
05 Application Session and service validation

This investigation model prevents engineers from confusing route visibility with actual forwarding. It also encourages systematic validation of return traffic, which is frequently overlooked in complex transit environments.

09.3 / Engineering Incident Explorer

Select a Production Failure Scenario

The following examples illustrate common AWS transit failures. Select an incident to review its symptoms, likely causes and recommended investigation.

Missing Transit Gateway Route

Symptom: Source VPC traffic cannot reach a destination VPC.

Likely cause: The ingress attachment's associated Transit Gateway route table lacks the required destination prefix.

Investigation: Verify source subnet routing, attachment association, destination route visibility and the reverse path.

09.4 / Routing Failure Matrix

Identifying Missing and Incorrect Routes

AWS routing failures can occur at several different forwarding boundaries. The correct troubleshooting approach depends on where the packet stops.

Failure Location Typical Symptom Validation
Source VPC route table Packet does not enter transit. Inspect destination route and target.
TGW route table association Unexpected routing-domain lookup. Identify ingress attachment's associated table.
TGW propagation / static routes Destination prefix absent or incorrect. Inspect effective route entries and attachment targets.
Cloud WAN segment policy Cross-segment connectivity missing or unintended. Inspect deployed policy and segment route visibility.
Inspection VPC Packet dropped or bypasses intended firewall. Trace endpoint routes, firewall logs and return path.
Hybrid routing On-premises prefixes missing or asymmetric. Inspect BGP, DX/VPN state and route filtering.
Destination VPC Traffic reaches destination network but session fails. Inspect subnet routing, SGs, NACLs and host state.
09.5 / Diagnostic Decision Tree

A Repeatable Fault-Isolation Workflow

AWS TRANSIT NETWORK TROUBLESHOOTING START: Define source IP, destination IP, protocol, port and expected path. QUESTION 1: Does the source subnet route point to the correct transit target? NO → Correct VPC routing. YES → Continue. QUESTION 2: Is the relevant attachment available and enabled for the required AZ? NO → Investigate attachment. YES → Continue. QUESTION 3: Does the effective transit route select the intended destination? NO → Investigate TGW route table or Cloud WAN segment policy. YES → Continue. QUESTION 4: Does traffic traverse the required firewall or inspection path? NO → Investigate traffic steering. YES → Continue. QUESTION 5: Does security policy permit traffic? NO → Review SG, NACL, firewall rules. YES → Continue. QUESTION 6: Is the return path valid? NO → Investigate reverse routing. YES → Continue. QUESTION 7: Is the destination application healthy? NO → Investigate application service. YES → Validate complete transaction. END: Document root cause, corrective action and evidence.
09.6 / Cloud WAN Policy Validation

Policy Intent Versus Effective Routing State

Cloud WAN introduces a declarative global routing model, but the intended policy and deployed routing state must be examined separately.

A core network policy may define a segment-sharing relationship, but a missing attachment, incorrect association or absent route can still prevent connectivity.

VALIDATION 01
Policy Deployment

Confirm the intended core network policy version was deployed successfully.

VALIDATION 02
Attachment Association

Verify attachments belong to the expected segments and regional core edges.

VALIDATION 03
Effective Route Visibility

Inspect route tables and segment-sharing behaviour to validate actual forwarding reachability.

09.7 / AWS Network Observability

Evidence-Driven Network Troubleshooting

Production troubleshooting should use available telemetry to distinguish packet forwarding, routing decisions, security drops and application failures.

Tool / Data Source Primary Purpose
VPC Flow Logs Observe supported network-interface traffic metadata.
Transit Gateway Flow Logs Analyse supported traffic flow metadata across TGW attachments.
VPC Reachability Analyzer Analyse supported network paths and configuration constraints.
Network Firewall Logs Investigate firewall alert and flow information.
CloudWatch Metrics Monitor supported networking service health and traffic indicators.
CloudTrail Review recorded AWS API and configuration activity.
Hybrid Router Logs Investigate BGP state, route changes and transport failures.

Observability principle: A flow log entry does not necessarily prove end-to-end application success. Correlate network telemetry with firewall decisions and application-level evidence.

09.8 / Production Incident Walkthrough

Case Study: Inter-VPC Connectivity Failure

INCIDENT: Production VPC cannot reach Shared Services VPC. SOURCE: 10.10.1.25 DESTINATION: 10.20.1.40 EXPECTED: TCP application connectivity. INITIAL CHECK: Source VPC route: 10.20.0.0/16 → Transit Gateway STATUS: PRESENT ATTACHMENT: Production attachment available. STATUS: HEALTHY TRANSIT ROUTING: Production attachment associated with Production-RT. ROUTE TABLE INSPECTION: 10.20.0.0/16 route missing. ROOT CAUSE: Shared Services attachment propagation not enabled into the required route table. REMEDIATION: Enable intended propagation or configure an appropriate route according to policy. VALIDATION: Confirm destination route appears. Validate reverse path. Test application connectivity. Review security policy. Record incident evidence.

This example demonstrates why attachment availability and VPC routing are insufficient to prove transit connectivity. The associated Transit Gateway route table must contain the intended destination route.

09.9 / Architecture Readiness Checklist

Production Validation Before Deployment

✓ Validate VPC CIDR planning and overlapping-prefix risks.
✓ Confirm Transit Gateway attachment AZ coverage.
✓ Review TGW route-table association and propagation.
✓ Validate Cloud WAN segments and deployed policy versions.
✓ Test permitted and prohibited network relationships.
✓ Confirm inspection path symmetry and firewall rules.
✓ Verify Direct Connect, VPN and BGP resilience.
✓ Test multi-region failure and recovery scenarios.
✓ Enable appropriate network logging and monitoring.
✓ Document recovery procedures, ownership and escalation.
09.10 / Engineering Guide Conclusion

From Transit Connectivity to Production Engineering

AWS Transit Gateway and AWS Cloud WAN provide complementary approaches to enterprise cloud networking. Transit Gateway delivers regional transit routing with attachment and route-table controls, while Cloud WAN provides a managed global core network with policy-driven segmentation and connectivity.

Production architectures must coordinate VPC routing, transit routing, segmentation, firewall inspection, hybrid connectivity and multi-region recovery.

The complete engineering model is: architecture design → route policy → attachment connectivity → security enforcement → traffic validation → incident investigation → resilience testing.

Engineering Guide 20 completed: AWS Transit Gateway and AWS Cloud WAN Architecture Explained — nine editorial sections covering regional transit, global core networking, segmentation, inspection, hybrid connectivity, resilience and production troubleshooting.

Network Insight · Engineering Guide 20 · Lab 09

AWS Network Incident Investigation Engine

Investigate production-style AWS networking incidents across Transit Gateway, Cloud WAN, VPC routing, inspection, and hybrid connectivity. Inspect control-plane evidence, identify the failure domain, apply repairs, and validate recovery using an interactive packet trace.

Production Network Investigation Topology

The architecture connects application and database workloads through Transit Gateway, a central inspection VPC, and hybrid enterprise connectivity.

AWS PRODUCTION NETWORK · INCIDENT TOPOLOGY APPLICATION VPC 10.10.0.0/16 Production workloads DATABASE VPC 10.20.0.0/16 Protected database tier TRANSIT GATEWAY Routing Policy ENTERPRISE NETWORK 172.16.0.0/16 Direct Connect / VPN INSPECTION VPC 10.40.0.0/16 Stateful Firewall
Transit Gateway Healthy path Failure domain
HIGH Incident severity
0 Repairs applied
OPEN Incident status
Select a network component to inspect its operational role.

Incident Response Engineering Console

Select an incident and begin the troubleshooting investigation.
Active Incident Description
Application workloads cannot reach the database network because a destination route is missing from the relevant Transit Gateway route table.

Control-Plane Evidence Inspector

Component Evidence Status

Incident Investigation Workflow

1. Identify the affected traffic flow
2. Inspect source and destination routing
3. Validate transit attachments
4. Evaluate security and segmentation
5. Identify root cause
6. Apply repair and verify recovery

Production Diagnostic Console

[NETWORK INSIGHT · INCIDENT LAB 09] AWS production network loaded. Incident: Missing TGW Route Severity: HIGH Status: OPEN Awaiting engineering investigation.

Root Cause and Recovery Analysis

Investigation objective: Identify the failure domain using route, attachment, policy, and packet-path evidence.
Recovery objective: Apply the appropriate configuration repair and validate the resulting traffic path.

Production Engineering Notes

AWS network troubleshooting requires distinguishing control-plane problems from data-plane failures. A healthy Transit Gateway attachment does not guarantee that the correct route is installed in the associated TGW route table or the VPC subnet route tables.

Cloud WAN segmentation failures may involve attachment association, segment policy, or route propagation. Centralised inspection failures may involve firewall policy, endpoint availability, asymmetric routing, or incorrect traffic steering.

Hybrid connectivity incidents should include Direct Connect and VPN health, BGP session state, advertised prefixes, Direct Connect gateway associations, route preference, and return-path validation.

This is a deterministic educational incident simulator. Evidence is generated from predefined scenarios rather than retrieved from AWS APIs. Repairs change the simulated state only and do not modify any real AWS infrastructure.

☕
Support Network Insight
Help support the development of interactive networking labs, BGP simulations, and educational content.
Support the Project
Matt Conran
Tags: No tags

Comments are closed.