Given zync db is ephemeral we need to know how 3scale upgrades or other scenarios that cause zync (db) to be unavailable affect the overall behaviour of 3scale and how it might recover (with the help of system).

Answer this question - do we need to externalize the zinc database, or not?

Something like this: Deploy 3scale, verify that zync is working, in between 2 requests, scale zync database deployment (pod) down to 0, keep working on 3scale, see what happens, scale zync database pod back to 1. See what happens. (if it fails, we advise clients to externalize database and do backups.)

Refs:
https://docs.google.com/spreadsheets/d/17jmES-_eEYqdCUxwUaTfmzp5G1zaDoGncKTegHJBMCA/edit?ts=5f313664#gid=0
https://docs.google.com/spreadsheets/d/1unjRoFO4cv3jo-kh5naXAAkPhEn5DnqFUmu7RxBeyeQ/edit#gid=2035812480