HomeArchitectureDistributed tracing brings deeper observability

Distributed tracing brings deeper observability

As organizations move towards service-oriented architectures and leave behind the monolithic workloads, they find themselves in uncharted territory. On the one hand, the microservices architecture makes development quicker and boosts collaboration among teams to deliver faster and meet SLAs. But, as workloads grow in size, they can become tough to monitor and observe. As more and more services are built and involved in a workload, it becomes hard to trace requests that can help teams test their applications.

Traditional monitoring tools are built to work with monolithic applications where request tracing is comparatively more straightforward. With the advent of microservices and distributed infrastructure, it has become harder to do the same using traditional tools as a request travel through several independent services in a dynamic environment where resources are leveraged and released continuously. So how can organizations trace these requests?

What is distributed tracing?

In modern applications, tracing isn’t a walk in the park. All the systems work together to execute user requests. Tracing a request is vital in distributed systems, even if efficient logging systems are in place. When different teams work on different services using various tech stacks and programming languages, it becomes incredibly daunting to identify services with performance issues.

When there’s no clear visibility into the distributed workloads, teams can turn on each other and play the blame game. At the same time, deliveries get delayed leading to an impact on the business. Traditionally, organizations would try to develop a tracing system tailor-made to their systems, but that’s too much work and can divert teams from what really matters.

This is where distributed tracing comes to the rescue. Distributed tracing is the process of observing a single transaction as it makes its way through several services inside a distributed application. Each request is tagged with a unique identifier that follows the request from the top of the stack through the application layer and then through various services. Metadata like service name and start and end timestamp, among other vital parameters, are collected as the activity aka span makes its way through a workload. Once the parent span is completed, the trace follows the child span, and a list of all the spans is created in the correct order so teams can spot errors and performance issues.

Benefits of distributed tracing

Distribute tracing helps organizations trace traditional logging metrics like timestamps and service information and identify key performance metrics that can help ITOps and DevOps teams. With distributed tracing, teams can:

  • Get stats on the health of the application and all the services within the said application.
  • Address performance issues and identify the responsible code quickly.
  • Collaborate with other teams to address issues within microservices quickly by delivering on SLAs and meeting customer satisfaction.
  • Improve user experience using information like success rates, latency, response time, and other user experience metrics.
  • Monitor transactions closely through dynamic dashboards.
  • Address issues arising due to autoscaling and upgrades before they can impact business.
  • Save time in testing and debugging and focus on innovation.

latest in distributed tracing

The latest in distributed tracing

As distributed tracing becomes more of a necessity, this space is booming with innovation. Observability tools are coming up with new and exciting features addressing business needs. Let’s take a look at recent developments in this space.

Grafana Tempo

Grafana Tempo is distributed tracing project that reached GA with v1.0 earlier this year. This tool helps teams store logs with trace ID and allows teams to search for relevant information without relying on an Elasticsearch or Cassandra cluster. Spans can be stored on the backend without needing indexing. This reduces I/O costs and makes it easy to query existing traces to identify issues and to make informed decisions to tackle them.

Grafana Tempo also compresses back-end traces and write-ahead logs and ensures distributed tracing doesn’t bog down system resources. Teams can also link Tempo with log data sources to get even deeper observability into their applications.


Thundra came out with Sidekick in 2021, a revolutionary debugging tool that debugs serverless applications running on remote platforms. This tool allows teams to debug running applications by implementing non-breaking breakpoints called tracepoints. These tracepoints capture the snapshot of traces at the point of tracepoint creation. Thundra Sidekick is armed with features like auto-instrumentation, auto distributed tracing, and remote debugging that help teams focus on their tasks.

Thundra’s testing product, Foresight, allows teams to test their application end-to-end or perform integration testing by combination tests with detailed information about the actions performed by each service inside the system. This feature allows teams to identify critical issues before the services are run in production.


OpenTelemetry is an open-source CNCF project. Previously in the sandbox stage, OpenTelemetry is now in the active development stage. OpenTelemetry is a combination of tools, APIs, and SDKs that help retrieve telemetry information like metrics, logs, and traces in order to help teams visualize requests inside their distributed applications. OpenTelemetry has the edge over paid distributed tracing solutions as an open-source and vendor-neutral solution. OpenTelemetry has the potential to become the standard tracing solution for service-oriented workloads.


Jaeger, another CNCF project, has a foot over Opentelemetry, being a graduated project. Jaeger comes packed with features like latency and performance optimization that help relay the tedious task of manually tuning distributed workloads. With root cause analysis, Jaeger helps identify performance issues automatically. Jaeger also performs service dependency analysis and distributed context propagation to help teams understand how their services work so they can have a bird’ eye view of their distributed workloads.


Distributed tracing is vital for organizations running service-oriented applications in production. With the insights it provides, teams can identify performance issues before users can notice them and helps organizations deliver an excellent user experience and ensure a steady flow of revenue.

If you have questions related to this topic, feel free to book a meeting with one of our solutions experts, mail to sales@amazic.com.


Receive our top stories directly in your inbox!

Sign up for our Newsletters