Guest: Austin Parker (LinkedIn)
Company: Honeycomb (Twitter)
Show: Let’s Talk

Austin Parker has moved from his role at Lightstep to Director of Open Source at Honeycomb. He was attracted to working with the company because he wanted to be part of a startup that he could watch scale and grow following the exit at Lightstep. Honeycomb has been instrumental in popularizing observability and innovating with OpenTelemetry and even though there are now new players in observability, Parker believes that they all look a lot like Honeycomb but it is a good sign if you are being copied.

“The OpenTelemetry project has achieved all the original goals that it set out to achieve five years ago, having last year announced that logging is stable and GA in the specification,” says Parker. There is more awareness of OpenTelemetry with companies of all sizes and even though they may not be sure what tools they will put on top of the stack, they are certain they are going to be using OpenTelemetry on the base layer. Many startups are also recognizing that they can start with open source and move into a proprietary solution like Honeycomb as they grow.

OpenTelemetry is evolving as it gets more adoption and more users. They need to listen to what people need in the community but a major challenge is controlling the scope and any changes they do make need to be carefully considered. They are focusing on two areas based on the community’s needs at the moment: real user monitoring and making OpenTelemetry more useful for web clients, iOS apps, and Android apps. Another area is continuous profiling and being able to get low-level data about what is going on in the code and attach it to everything else in the OpenTelemetry ecosystem.

Honeycomb recently announced Honeycomb for Kubernetes which is unique since it emits so much telemetry data due to its complexity. The solution was created to help applications understand what is happening in Kubernetes so they can gain a better understanding of how things such as memory pressure on a node, over or under schedules, and overprovisioning may be impacting their customer experience. Honeycomb for Kubernetes provides the ability to pull a curated set of metrics, logs, and traces out of Kubernetes clusters to help developers see the relationship between their app and Kubernetes performance.

Whereas six or seven years ago, lots of companies were coming out with Kubernetes installers. Nowadays people who are running Kubernetes on their own hardware tend to use something that provides more than just Kubernetes. Nonetheless, the application can be installed on OpenShift, DigitalOcean, Google Kubernetes Engine, or others because API has become the standard. In the future, APIs are expected to become more standardized and built into more things to interact with telemetry subsystems or container orchestration subsystems. However, it is expected that the data model will form the standard default that people use to express these telemetry points and that is going to require changes in the market.

This summary was written by Emily Nicholls.