No video of the event yet, sorry!

This talk discusses the Four Golden Signals from the Google SRE handbook in the context of a Flask/Django/FastAPI/Go/ExpressJS/Spring Boot app. The signals are latency, traffic, errors and saturation. For web apps:

Latency is the response time for requests. At a high level, this can be the percentiles (50, 95, 99, 99.9 etc) and this can also be broken down by path or groups of paths. Traffic is the number of requests reaching the app. A total number in a given period of time (second, minute, hour etc) is usually interesting and this can also be broken down by path. Errors are any 4XX and 5XX responses from your app. 4XX can be interesting because they indicate that your users don’t know how to use your app. 5XX are even more interesting because they indicate problems in the code of the app. Saturation is an indication of how much of the total capacity of the deployment is being used. Once saturated, you would expect to see higher latency or requests starting to time out. Saturation can be indicated by high CPU or memory utilisation and web server worker idle time.

These metrics can be gathered from your app and then sent to tools like prometheus and visualised using grafana. We will then look at tooling from Canonical which provides builtin support for observability.