Workshop: What’s the future of o11y for serverless?
Last week, we had the pleasure of hosting a workshop on "the future of observability for serverless". It was quite exciting to have 3 AWS serverless heroes, a Microsoft MVP, a Serverless Java SME, a serverless meetup organiser and multiple AWS Community builders in the same room participating in this discussion.
During the workshop, we talked about the challenges teams currently face when it comes to observability for serverless, and the opportunities to overcome them.
One of the main challenges we discussed was how difficult it is to create a culture of observability within teams. This goes from enabling every engineer to define meaningful alerts and SLOs, to providing tooling with a pricing model that incentivises better observability, in order to drive better business outcomes ultimately.
Questions around observing “codeless” service integrations also came up, as these can be particularly difficult to debug when there’s a defect.
Many ideas to overcome these challenges were suggested. Generative AI was mentioned of course, but it’s no silver bullet. The first steps would be centered around making instrumentation as easy and consistent as possible for developers and building an experience that adapts to the software architecture such that signals rise to the top when making use of the telemetry data. As Sheen Brisals from the LEGO Group put it: “Observing activities in a pond is way more manageable than in an ocean!”.
We also discussed the current disconnect between X-Ray and OpenTelemetry, and how distributed tracing remains difficult within a serverless environment without severely impacting performance. A lot of work remains to be done on this front.
More importantly, throughout the workshop, we had a great time networking with the community and learning from each other's experiences. I was impressed with the enthusiasm and expertise of everyone in the room and how open the conversations were. It is clear that the future of observability for serverless is bright, and that there is a lot of exciting work ahead!
This is just the beginning, there’s a lot of room for innovation in the serverless observability space. I’d love to hear from you to continue exploring these ideas and to share experiences with the community!
👋 Say Goodbye to Confusing Error Messages with Baselime AI
Baselime AI simplifies debugging by providing clear explanations for any chart, log, event, metric or trace.
The Serverless Guide to AWS CloudWatch Logs Insights
Demistifying your system's deepest secrets with Cloudwatch Log Insights and sharing its biggest gotcha. This blog post will make you a Cloudwatch Log Insight pro in just 10 minutes.
Service discovery, dashboards, reliability and much more...