On Thu, Mar 6, 2025, 7:32 AM Greg Sabino Mullane <htamfids@gmail.com> wrote:
On Wed, Mar 5, 2025 at 9:44 PM me nefcanto <sn.1361@gmail.com> wrote:
Anyway, that's why I asked you guys. However, encouraging me to go back to monolith without giving solutions on how to scale, is not helping.
We did. In addition to the ongoing FDW discussion, I mentioned read-only replicas and Citus. As far as *how* to scale vertically, we can offer general advice (more hardware resources, ramdisks for temp stuff, OS-level tuning, separate disk mounts). But a lot of it is tuning Postgres for your specific situation and your specific bottlenecks. Which we are happy to help with. Once we convince you to not throw the baby out with the bathwater. :)
8 Terabytes of data. A single backup took us more than 21 days
Something was fundamentally wrong there.
It could happen on an old and drained hardware... 😀
8TB databases existed 20+ years ago. Like always, the hardware must fit the application.
21 days to backup a database absolutely means many things were improperly sized and configured.
