We recently migrated our Looker backend database from the in-memory HyperSQL setup to an instance of MySQL running in AWS RDS. We weren't experiencing any performance issues with HyperSQL, but we wanted to go ahead and make the switch not only for eventual demand, but also for easier backup, DR planning, more selective migration of data between environments, etc etc.
I configured the MySQL instance according to this Discourse post:
Works great. Then we got the bill. Yikes.
So now I'm taking a look at what we can scale down without running into problems. There are 3 settings I think could be changed
- Multiple availability zones: Yes
- Storage Type: Provisional IOPS
- Class: db.m3.medium
Multi-AZ is probably a no-brainer. I can turn this off and it will probably save us quite a bit.
I'm less certain about the storage type. Our dev instance is using General Purpose and it seems to work fine. But I don't know if this would be less performant in a Prod environment.
Same goes for the machine class. I hesitate to lower this without knowing what effects it might have on a production instance.
Has anyone tried scaling these settings down in production? Any recommendations?
Thanks in advance