2 ms·
> Just snapshot the whole VM if you need a backup. How do you deal with things like updates? Upgrades to your service? Do you just accept that you have schedu
by gwbas1c 6y ago
> Just snapshot the whole VM if you need a backup.
How do you deal with things like updates? Upgrades to your service?
Do you just accept that you have scheduled downtime when your service won't be available?
- bob1029 6y agoYes - We have allowed maintenance windows each day. That said, we have prototypes of architectures in which we have multiple instances running in the same production environment simultaneously, each with an independent SQLite database and some light-weight replication logic in the application itself. Simple DNS RR or customer-managed LB would be responsible for routing client traffic. There aren't a whole lot of entities that we actually need to synchronously replicate between application servers, so this is far more accessible for us to iterate on than throwing our hands up and jumping to some fully-managed always-on clustered database service and throwing away all of the lessons we've learned with SQLite.