Is there a notable difference between HA offered b...
# general
s
Is there a notable difference between HA offered by Postgres SaaS providers vs. doing it yourself? In other words, are the SaaS providers using some proprietary software/patches that work better than the pure open source tools? I’m hearing it’s difficult to set up HA Postgres and that it frequently requires some manual operations
l
I think it just comes down to how much time you want to spend managing the database. The SaaS providers manage it for you but if it's down there's not much you can do. If you manage it yourself you are at the mercy of your own mistakes
Most people choose not to manage it themselves so they can focus on building their product
c
One of the core value props of Yugabyte's open-source postgres-compatible software (https://www.yugabyte.com/) is that it makes it easier to do an HA deployment than with vanilla postgres.
(Disclaimer: I'm personally a very small investor in YB and am friends with some of the people there—they're good people working on good tech)
g
I don't think its hard, and the OSS tools are great - for HA. EDB have proprietary tools that are considered far better than the OSS ones. I think the big deal is that they have active-active. AWS has pure OSS version (RDS) and proprietary stack (Aurora). Not sure about everyone else.
@Colt McNealy I didn't realize that you invest in a competitor 🙂 I need to be careful around you now!
s
Why do people seem to say it’s hard? Is something missing from these like easily telling when the master is dead? Or are the db admins just not super used to distributed systems ?
I’m doing something interesting and unconventional which is comparing Postgres as a queue and pub sub versus Kafka :)
I’m surprised by how much PG scales with sync commit, considering it uses fsync too