About replication - does DB replication provide any advantage, compared to RAID1 with the same number of replicas? I think that with RAID1, if sector 100 of disk1 is unreadable, and sector 200 of disk2 is unreadable, then you still have all the data and it is easy to recover. By contrast, with DB replication I don't see a way to recover, at least haven't found anything in Postgres documentation. So RAID1 seems safer. Am I missing something? Maybe DB replication is more about availability if HW (not just disk) breaks down?
postgreSQL db replication can be done across machines and data-centers--that's more of a cloud-centric model for scaling and redundancy, but if you want to be as production grade as a social network or financial institution, it might be desired to have the option to do offsite live replication. You can setup postgres in active-active so you could have a failsafe node that boots up and uses a remote copy of the db and just takes over... this is all much more detailed though for a pleb. Mostly, the advice for the node runner starting out is to do some reseach and pick a technology stack that is going to be the direction of least pain for studying and implementing, but in reality, any and all of the options can blow up.
reply