We use both average and quantiles to measure latencies, but monitoring tools often have limitations with percentiles, which sometimes produce counterintuitive results. While investigating a latency problem, we noticed that Prometheus -- the system we use for monitoring -- reported the P99 latency for our app was less than the

Observability of modern cloud-native applications should provide deep visibility into an application’s performance, call attention to problems brewing through high fidelity alerts and provide an intuitive troubleshooting experience that shouldn’t require expert domain knowledge. With the advent of open-source approaches either based on OpenTelemetry or the ubiquitous Prometheus

Compared to the old term “monitoring”, the new buzzword “observability” is not merely a marketing slogan. It does convey a new set of challenges with the cloud-native paradigm shift. This blog provides an engineer’s perspective on these challenges and why Asserts thinks Knowledge Graph has a lot of potential

IntroductionSpringboot is a popular Java framework for building modern applications. With the help of the actuator module and the micrometer library, we can configure a Springboot application to expose performance metrics in the Prometheus format. Then we can set up a monitoring backend to consume these metrics. This blog provides

AWS LambdaAWS Lambda allows you to write functions and run them as a scalable service. It supports popular languages and runtimes like JavaScript/NodeJS, Python, Go and Java. It allows an event-driven model wherein Lambda functions are invoked in response to events, e.g., requests to an API Gateway, messages

IntroductionPrometheus has four metric types, among which the two basic ones are counter and gauge. A gauge is intuitive as it matches our expectation for a metric, i.e., it represents a numeric value that measures something like temperature, memory usage, etc. A counter that monotonically increases, on the other

In the previous post in this series, I explained why alerts should be based on the user-visible symptoms of problems rather than on problem causes. A quick recap: it is extremely hard to reliably alert on all the possible failure modes of a complex system, so a cause-based alerting strategy

If you follow application monitoring and the numerous monitoring tools vying for developer attention, you've probably noticed that everyone seems to be talking about service level objectives (or SLOs) all of a sudden. There are plenty of resources on the web defining what SLOs are – if you haven't seen them,