Thank you guys for your inputs and suggestions.
I completely agree that zero downtime is not 100% achievable hence distributed database solutions are quite popular too .
Thanks Deepak
From: akshay polji <akshay.polji@gmail.com>
Sent: Wednesday, May 1, 2024 1:08:48 AM
To: Scott Ribe <scott_ribe@elevated-dev.com>
Cc: pgsql-admin <pgsql-admin@postgresql.org>
Subject: Re: HA Setup Review
"All cloud providers have downtime and outages. Maybe this is reliable enough for you, but just don't treat it as a guarantee." --> Very true!!
"if a network glitch, or failure of a replica, can cascade to cause failure of primary, is that really HA?" --> Absolutely not.
But then for HA you need a replica, so now the question is if you want to do Async or Sync.
Sync would mean impact to Primary due to n/w issues or failures but guarantee no-to-minimum data loss,
Async would mean no impact to Primary due to n/w issues or failures but can cause data loss due to replication lag.
If you ask the applications, they will need both .. Neither impact to primary nor data loss. How to weigh the options in this case?
Unless there's a middle ground that I am not aware of but would love to know more.
Thanks,
Akshay
> On Apr 30, 2024, at 10:29 AM, akshay polji <akshay.polji@gmail.com> wrote:
>
> So do you think that even such a cluster with 3 node pgpool + postgresql (running on the same machine) Synchronous Replication (Any one out of the two replicas) would mean Primary DB will be at risk of degraded performance?
All cloud providers have downtime and outages. Maybe this is reliable enough for you, but just don't treat it as a guarantee. The point is: if a network glitch, or failure of a replica, can cascade to cause failure of primary, is that really HA?