Why code instrumentation?
I am writing this blog post as a common introduction for a new category I would like to write about consistently on my blog. If this is the first time you land here this is my blog, and I write about everything that catches my attention but sometime earlier sometime later on I realize I can group my posts in categories, and that’s what I am doing now.
There are a lot of people writing about observability, monitoring and I did it for the last 3 years as well. I learned a lot along the way but what I think is crucial is that developers has to write code that is understandable and easy to debug where it is more valuable, in production. And if an application or a system is hard to figure out we as a developer play a mojor role on it.
That’s why Site Reliability Engineering (SRE) is not related to ops, servers, Kubernetes but it is something that plays its match in your code.
What’s why I think SRE and DevOps are different, not at all connected.
The technologies that are leading the landscape are:
- Prometheus but not the time series database, their client libraries and the exposition format, now branded from the community and the Cloud Native Computing Foundation (CNCF) as OpenMetrics
- OpenTracing, OpenCensus and OpenTelemetry. They are part of the same bullet points because I think about them as the consequence of each other since what I hope is “THE LAST ONE”, OpenTelemetry. They are instrumentation libraries and specification to increase interoperability and to avoid vendor lock-in for what concerns distributed tracing and metrics. I hope logs will jump onboard at some point
Prometheus and OpenMetrics
I wrote about this topic previously, so have a look there if you do not know what I am speaking about.
I think they are worth to mention here because that’s how I learned the effect of good or bad code instrumentation, and the fact that it has to happen in your code, when you develop it.
It has the same weight has writing a good data structure, writing solid unit tests, or picking the right design pattern.
As I said I will refer to otel when I can, not because I think OpenTracing or OpenCensus is bad, but because I do not see this as a religion, but for me it is a technical problem, they is well spread and it has to find a good answer.
Those communities decided to merge to otel in the way they are doing, good or bad? We can get a beer at some point and I will tell you. It is out of scope.
What I am gonna talk about
This is a long new category introduction blog post probably but that’s it. Over the last two years I tried to share what I experienced around this topic with a workshop called: “Application Monitoring”. A lot of the articles that I will write comes from there, and it is an attempt to share what I think worked or failed.
- All about Code Instrumentation from my blog
- ShopMany is the application I developed for the workshop
- Workshop notes contains notes, exercises and solutions for the lessons I proposed in the workshop itself
- honeycomb because when you speak about o11y you have to quote them!
- My newsletter is probably the best way to stay in touch with the content I create
- Twitter is the best way to stay in touch with me