Services should surface the information required to prevent throttling.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | cmassaro | T394917 If we understand capacity planning, we can create an architectural plan how to prevent service outage/throttling, to be executed in subsequent quarters | |||
| Resolved | cmassaro | T399318 Write an architectural plan for capacity planning work in a subsequent quarter | |||
| Resolved | cmassaro | T399317 Learn what "capacity planning" means at WMF |
I 've gone ahead and shaped up some older notes I had in wikitech and posted a guide for capacity planning in https://wikitech.wikimedia.org/wiki/Kubernetes/Capacity_Planning_of_a_Service
Have a go at eat and let me know if something feels off, doesn't add up, isn't clear etc.
It's generic enough to apply to any stateless workload so it should work out for wikifunctions.