In the past we've had quite a bit of churn of SLO definitions and especially the implementation of of monitoring/alerting for them.

We should decide two broad topics:

• what the usual SLIs are, and how to determine the SLOs (thresholds/quantiles/...) that they then are associated with
• how new services' SLOs are put into the new framework (Sloth, see https://wikitech.wikimedia.org/wiki/Sloth )

This then should inform a re-work where necessary of existing SLOs/services as well as making a generic guide (WT page) on how LW SLOs are decided and created.