observability platform

Observability vs Monitoring

To understand the difference between observability vs monitoring, we need to first discuss the role of monitoring. Monitoring is the evaluation to help identify the most valuable and efficient use of resources. So the big question I put to you is what to monitor. This is the first step to preparing a monitoring strategy and there are a couple of questions you can ask yourself to understand fully if monitoring by itself is enough or do you need to move to an observability platform. Firstly, you should consider what you should be monitoring, why you should be monitoring, and how to monitor it.?  When you know this, you can move into the different tools and platforms available. Some of these tools will be open source and others commercial. When evaluating these tools, one word of caution: does each tool work in a silo, or can it be used across technical domains? Silos are breaking agility in every form of technology.

 

Observability vs Monitoring

Diagram: Observability vs Monitoring

 

The Starting Point: Observability vs Monitoring

You need to measure and gather the correct event information in your environments, and this will be done with several tools. This will let you know what is affecting your application performance and infrastructure. As a good starting point, there are four golden signals to look out for: There is latency, saturation, traffic, and errors. These are Google’s Four Golden Signals. The four most important metrics to keep track of are: 

      1. Latency: How long it takes to serve a request
      2. Traffic: The number of requests being made.
      3. Errors: The rate of failing requests. 
      4. Saturation: How utilized the service is.

So now we have some kind of a guide on what to monitor, and let us apply this to Kubernetes to, for example, let’s say a frontend web service that is part of a tiered application, we would be looking at the following:

      1. How many requests is the front end processing at a particular point in time,
      2. How many 500 errors are users of the service received, and 
      3. Is the service overutilized by request?

So we already know that monitoring is a form of evaluation to help identify the most valuable and efficient use of resources. With monitoring, we observe and check the progress or quality of something over time. So within this, we have metrics, logs, and alerts. Each has a different role and purpose.

 

    • Monitoring: The Role of Metrics

Metrics are related to some entity and allow you to view how many resources you consume. The metric data consists of numeric values instead of unstructured text such as documents and web pages. Metric data is typically also time series, where values or measures are recorded over some time.  An example of such metrics would be available bandwidth and latency. It is important to understand baseline values. Without a baseline, you will not know if something is happening out of the norm. What are the usual baseline values of the different metrics for bandwidth and latency? Are there any fluctuations in these metrics? How do these values rise and fall during normal operations and peak usage? And this may change over the different days in the week and months. If, during normal operations, you notice a rise in these values. This would be deemed abnormal and should act as a trigger that something could be wrong and needs to be investigated. Keep in mind that these values should not be gathered as a once-off and can be gathered over time to give you a good understanding of your application and its underlying infrastructure.

 

    • Monitoring: The Role of Logs

Logging is an essential part of troubleshooting application and infrastructure performance. Logs give you additional information about the events. This is important for troubleshooting or discovering the root cause of the events. Logs will have a lot more detail than metrics. So you will need some way to parse the logs or use a log shipper. A typical log shipper will take these logs from the standard out in a Docker container and ship them to a backend for processing. FluentD or Logstash has its pros and cons and can be used here to the group and sent to a backend database that could be the ELK stack ( Elastic Search). When using this approach, you can add different things to logs before sending them to the backend. For example, you can add GEO IP information. And this will add richer information to the logs that can help you troubleshoot.

 

    • Monitoring: The Role of Alerting

Then we have the alerting, and it would be best to balance how you monitor and what you alert on. So, we know that alerting is not always perfect, and it will take time to get the right alerting strategy in place. It’s not a simple day one installation and requires much effort and cross-team collaboration. You know that alerting on too much will cause alert fatigue. And we are all too familiar with the problems alert fatigue can bring and tensions to departments. To minimize this, you need to consider Service Level Objective (SLO) for alerts. SLOs are measurable characteristics such as availability, throughput, frequency, and response times. Service Level Objectives are the foundation for a reliability stack. Also, it would be best if you also considered alert thresholds. If these are too short, you will get a lot of false positives on your alerts. 

Even with all of these in place, monitoring is not enough. Due to the sheer complexity of today’s landscape, you need to consider and think differently about the tools you use and how you use the intelligence and data that you receive from these tools to resolve issues before they become incidents.  Like that, monitoring by itself is not enough. The tool used to monitor is just a tool that probably does not cross technical domains, and there will be different groups of users who will administer each tool without a holistic view. The tools alone can take you only half the way through the journey.  Also, what needs to be addressed is the culture and the traditional way of working in silos. A siloed environment can affect the monitoring strategy you want to implement. Here you can look into an Observability platform.

 

Observability vs Monitoring

Diagram: Observability vs Monitoring. Link to YouTube video.

 

Observability vs Monitoring

So when it comes to observability vs monitoring we know that monitoring can detect problems and tell you if a system is down and when your system is UP, Monitoring doesn’t care. Monitoring only cares when there is a problem. The problem has to happen before monitoring takes action. It’s very reactive. So if everything is working, monitoring doesn’t care. On the other hand, we have an Observability platform, a more proactive practice. It’s about what and how your system and services are doing. Observability lets you improve your insight into how complex systems are working, and let’s quickly get to the root cause of any problem known and unknown. Observability is best suited for interrogating systems to explicitly discover the source of any problem, along any dimension or combination of dimensions, without first predicting. This is a proactive approach.

 

The Pillars of Observability

This is achieved by utilizing a combination of logs, metrics, and traces. So we need to have data collection, storage, and analysis across these domains. While also being able to perform alerting on what matters most. Let’s say you want to draw correlations between units like TCP/IP packets and HTTP errors experienced by your app. The Observability platform pulls the context from different sources of information like logs, metrics, events, and traces into one central context. Distributed tracing adds a lot of value here. Also, when everything is placed into one context, you can switch between the necessary views to troubleshoot the root cause accordingly easily. A good key component of any observability system is to have the ability to view these telemetry sources with one single pane of glass. 

Distributed Tracing in Microservices

Diagram: Distributed Tracing in Microservices

 

Known and Unknown / Observability Unknown and Unknown

Monitoring automatically reports whether known failure conditions are occurring or are about to occur. In other words, they are optimized for reporting on unknown conditions about known failure modes. This is referred to as known unknowns. In contrast, Observability is centered around discovering if and why previously unknown failure modes may be occurring: in other words, to discover unknown unknowns.

The monitoring-based approach of using metrics and dashboards is an investigative practice that leads with the experience and intuition of humans to detect and make sense of system issues. This is ok for a simple legacy system that fails in predictable ways, but the instinctual technique falls short for modern systems that fail in very unpredictable ways. With modern applications, the complexity and scale of their underlying systems quickly make that approach unattainable, and we can’t rely on hunches. Observability tools differ from traditional monitoring tools because they enable engineers to investigate any system, no matter how complex. You don’t need to react to a hunch or have intimate system knowledge to generate a hunch.

 

Observability and Controllability

Diagram: Distributed Tracing Explained: Link to YouTube video.

 

  • Monitoring vs Observability: Working Together?

Monitoring best helps engineers understand infrastructure concerns. While Observability best helps engineers understand software concerns. So Observability and Monitoring can work together. First, the infrastructure does not change too often, and when it fails, it will fail in more predictable ways. So we can use monitoring here. This is in comparison to software system states that change daily and are not predictable. Observability fits this purpose. The conditions that affect infrastructure health change infrequently, relatively easier to predict. We have several well-established practices to predict, such as capacity planning and the ability to remediate automatically (e.g., as auto-scaling in a Kubernetes environment. All of which can be used to tackle these types of known issues. Due to its relatively predictable and slowly changing nature, the aggregated metrics approach monitors and alerts perfectly for infrastructure problems. So here, a metric-based system works well. Metrics-based systems and their associated alerts help you see when capacity limits or known error conditions of underlying systems are being reached. Now we need to look at monitoring the Software. Now we need access to high-cardinality fields. This may include the user id or a shopping cart id. Code that is well-instrumented for Observability allows you to answer complex questions that are easy to miss when examining aggregate performance.

 

System Observability

System Observability: The Different Demands

We have had a considerable drive with innovation that has spawned several megatrends that have affected how we manage and view our network infrastructure and the need for observability. In reality, we have seen the decomposition of everything, from one to many. Many services and dependencies in multiple locations need to be managed and operated instead of the monolithic where everything is generally housed internally. The megatrends have resulted in a dynamic infrastructure with new failure modes not seen in the monolithic forcing us to look a different system observability tools and practices.

There has also been a shift in point of control. We move towards new technologies, and many of these loosely coupled services or infrastructure your services lay upon are not directly under your control. The edge of control has been pushed, creating different types of network and security perimeters. These parameters are now closer to the workload than a central security stack. Therefore the workloads themselves are concerned with security. For a more detailed explanation of these changes that drive the need for good observability and how they may effect you, a full 2-hour course I did for Pluralsight on DevOps Operational Strategies can be found here: DevOps: Operational Strategies.

 

System Observability Design

Diagram: System Observability Design.

 

  • How This Affects Failures

The major issue that I have seen with my clients is that application failures are no longer predictable, and the dynamic systems can fail in very creative ways challenging existing monitoring solutions. But, more importantly, the practices that support them. We have a lot of partial failures that are not just unexpected but not known or ever seen before. For example, if you recall, we have the network hero. 

 

  • The Network Hero

It is someone that knows every part of the network and has seen every failure at least once before. They are no longer useful in today’s world, and you need proper Observability. When I was working as an Engineer, we would have plenty of failures, but more than likely, we would have seen them before. And there was a system in place to fix the error. Today’s environment is much different. We can no longer rely on simply seeing either a UP or Down and setting static thresholds and then alerting based on those thresholds. A key point to note at this stage is that none of these thresholds consider the customer’s perspective.  If your POD is running at 80% CPU, does that mean the customer is unhappy? When looking to monitor, you should look from your customer’s perspective and what matters to them. Content Delivery Network (CDN) was one of the first to realize this game and measure what matters most to the customer.

 

The Different Demands

So the new modern and complex distributed systems place very different demands on your infrastructure and the people that manage the infrastructure.  For example, in microservices, there can be several problems with a particular microservice:

    • The microservices could be running under high resource utilization and therefore slow to respond, causing a timeout
    • The microservices could have crashed or been stopped and is therefore unavailable
    • The microservices could be fine, but there could be slow-running database queries.
    • So we have a lot of partial failures. 

 

Therefore: We Can No Longer Predict

The big shift we see with software platforms is that they are evolving much quicker than products and paradigms that we are using, for example, to monitor them. As a result, we need to consider new practices and technologies with dedicated platform teams along with good system observability. We really can’t predict anything anymore, which puts the brakes on some traditional monitoring approaches, especially the metrics-based approach to monitoring. I’m not saying that these monitoring tools are not doing what you want them to do. But, they work in a siloed environment, and there is a lack of connectivity. So we have monitoring tools working in silos in different parts of the organization and more than likely managed by different people trying to monitor a very dispersed application with multiple components and services in various places. 

 

Diagram: Prometheus Monitoring: Link to YouTube video.

 

  • Relying On Known Failures: Metric-Based Approach

A metrics-based monitoring approach relies on having encountered known failure modes in the past. The metric-based approach relies on known failures and predictable failure modes. So we have predictable thresholds that someone is considered to experience abnormal.  Monitoring can detect when these systems are either over or under the predictable thresholds that are previously set. And then, we can set alerts and hope that these alerts are actionable. This is only useful for variants of predictable failure modes.  Traditional metrics and monitoring tools can tell you any performance spikes or notice that a problem occurs. But they don’t let you dig into the source of problems and let us slice and dice or see correlations between errors. If the system is complex, this approach is harder to get to the root cause in a reasonable timeframe.

With the traditional style metrics systems, you had to define custom metrics, and these were always defined upfront. So with this approach, we can’t start to ask new questions about problems. So it would be best if you defined the questions to ask upfront. Then we set performance thresholds and pronounce them “good” or “bad.” and check and re-check those thresholds. We would tweak the thresholds over time, but that was about it. This monitoring style has been the de facto approach, but we don’t now want to predict how a system can fail. Always looking and always observing instead of waiting for problems, such as reaching a certain threshold before acting.

System Observability Analysis

Diagram: System Observability Analysis.

 

  • Metrics: Lack of Connective Event

Metrics did not retain the connective event. As a result, you cannot ask new questions in the existing dataset.  These traditional system metrics could miss unexpected failure modes in complex distributed systems. Also, the condition detected via system metrics might be unrelated to what is happening. An example of this could be an abnormal number of running threads on one component that might indicate garbage collection is in progress. It might also indicate that slow response times might be imminent in an upstream service.

 

  • Users Experience: Static Thresholds

User experience means different things to different sets of users. We now have a model where different service users may be routed through the system in different ways, using different components, providing experiences that can vary widely. We also know now that the services no longer tend to break in the same few predictable ways over and over.  We should have few alerts triggered by only focusing on symptoms that directly impact user experience and not because a threshold was reached.

 

  • The Challenge: Can’t Reliably Indicate Any Issues With User Experience

If you are using static thresholds, they can’t reliably indicate any issues with user experience. Alerts should be set up to detect failures that impact user experience. Traditional monitoring falls short in trying to do this. With the traditional metrics-based monitoring, we rely on static thresholds to define optimal system conditions, which has nothing to do with user experience. However, modern systems change shape dynamically under different workloads. Static thresholds for monitoring can’t reflect impacts on user experience. They lack context and are too coarse.

 

The Need For System Observability

System observability is a practice. Rather than just focusing on a tool that does logging, metrics, or altering, Observability is all about how you approach problems, and for this, you need to look at your culture.  So you could say that Observability is a cultural practice and allows you to be proactive to findings instead of relying on a reactive approach that we are used to in the past.  Nowadays, we need a different viewpoint, and we generally want to see everything from one place. You want to know how the application works and how it interacts with the other infrastructure components, such as the underlying servers, physical or server, the network, and how data transfer looks in a transfer and stale state.  What level of observation do you need so you know that everything is performing as it should? And what should you be looking at to get this level of detail?

Monitoring is knowing the data points and the entities we are gathering from. On the other hand, Observability is like when you put all of the data together. So monitoring is the act of collecting data, and Observability is putting it all together in one single pane of glass. Observability is observing the different patterns and deviations from baseline; monitoring is getting the data and putting it into the systems.

 

The 3 Pillars of Observability

We have three pillars of System Observability. There are Metrics, Traces, and Logging. So it is an oversimplification to define or view Observability as just having these pillars. But for Observability, you need these in place. Observability is all about how you connect the dots from each of these pillars. If someone asked me which one I prefer, it would be distributed tracing. Distributed tracing allows you to visualize each step in service request executions. As a result, it doesn’t matter if services have complex dependencies. You could say that the complexity of the Dynamic systems is abstracted with distributed tracing.

 

Distributed tracing

Diagram: Distributed Tracing: Link to YouTube video.

 

  • Use Case: Challenges Without Tracing

For example, latency can stack up if a downstream database service experiences performance bottlenecks. As a result, the end-to-end latency is high. By the time that latency is detected three or four layers upstream, it can be incredibly difficult to identify which component of the system is the root of the problem because now that same latency is being seen in dozens of other services.

 

  • Distributed Tracing: A Winning Formula

Modern distributed systems tend to scale into a tangled knot of dependencies. Therefore, distributed tracing shows the relationships between various services and components in a distributed system. Traces help you understand system interdependencies. Unfortunately, those inter-dependencies can obscure problems and make them particularly difficult to debug unless their relationships are clearly understood.