Value of your metrics
Goodbye monitoring, welcome observability
@gianarb - gianarb.it
Observability vs Monitoring
Learning vs eye examination
Debugging vs Ops work
SWEs should run their code
We need Ops people. I Ops
● Event
● Metric
● Tracing
Terms:
Logs are just a transport
layer
Instrumentation is part of the
development workflow
dashboards
you
Staging environment are useless
Allow your team to test in production
Why all this troubles?
● Cloud
● Containers
● Distributed systems
Build tools to have a better
point of view on your system
Do you need more?
● @mipsytipsy Charity Major on twitter
● @gianarb (it’s my twitter username)
● https://gianarb.it/blog/observability
● https://www.youtube.com/watch?v=1wj
ovFSCGhE

Value of your metrics: goodbye monitoring, welcome observability

Editor's Notes

  • #3 Observability is more around how to learn a codebase. It is about debugging not about dashboards or graphs. When you get paged and you look at one of your dashboards you troubleshoot quick because you already know what the problem is. Dashboard don’t speak enough for somebody that doesn’t know the app.
  • #4 I am Software Engineer, deploy my code to production is the best experience I never had and it took me to the next level. Production environment is the best learning experience for a developer to understand how its code really works or does not with real traffic, real customers but you need to give them the right tools and the right skills to build a comfortable production environment
  • #5 All of them are useful
  • #6 Instrument application for logs, metrics, traces can be expensive
  • #7 You build dashboard from answer that you had after a page. A lot of pages, a lot of dashbords... It doesn’t scale and it doesn’t look a work for engineers
  • #8 Distributed system are complicated. Replicate them is hard and keep databases in sync, reproduce traffic is hard.. So the best place to test is production.