GitLab Support insight into customer GitLab metrics
We could optimize much of our support engagements surrounding operating the GitLab service, by being able to have quick and easy insight into the Prometheus metrics of the instance. Instead of having to ask for logs, go back once or twice for additional data, etc. a support engineer could simply pull up the metrics and have a much more educated understanding of how their instance is performing.
This would also help our SA's and other pre-sales engagements, helping to tune and assist in getting GitLab up and running in more complex environments.
Looking even further, we could potentially derive some learnings out of this additional data:
- What is the average response time across instances?
- What are the common issues customers are running into, how can we avoid them?
- We could compare performance between releases across a wide array of environments
- We also could build and maintain alerts and other rules to pro-actively notify customers and support of issues. These could also be updated centrally and at once as opposed to requiring an update of the GitLab instance on prem.
- ... and plenty more.
We could support this by running a multi-tenant cloud based Prometheus service (perhaps based on Weave Cortex. Data gathering would be optional, of course, and filtered of any potential sensitive metrics that may be gathered in the future. (Project stats, etc)