Sample Header Ad - 728x90

Does TimescaleDB/PostgreSQL work well with Azure Disk backup?

0 votes
0 answers
81 views
I'm investigating improvements to a self-hosted TimescaleDB in Microsoft Azure. We have Timescale running in an AKS cluster, with a Standard SSD Azure Disk mounted as its persistent storage. Thus far we've used pg_dump for our backup strategy, but we're dissatisfied with its running time and the dump size. Our research suggests that Azure Disk backup may be appropriate for our use case. [Microsoft claims](https://learn.microsoft.com/en-us/azure/backup/disk-backup-overview?source=recommendations#key-benefits-of-disk-backup) that Azure Disk backup is useful when "invoking freeze and thaw on Linux virtual machines to get application-consistent backup puts undue overhead on production workload availability. To the best of our understanding, that is indeed how one would build TimescaleDB filesystem backups from the ground up. [The PostgreSQL documentation](https://www.postgresql.org/docs/current/continuous-archiving.html) also notes that "we can combine a file-system-level backup with backup of the WAL files...We do not need a perfectly consistent file system backup as the starting point." This also bodes well, but it is perhaps a bit unclear how far from "perfectly consistent" we can reasonably get before things break. Is Azure Disk backup the sort of atomic operation that would help prevent an inconsistent/corrupted database? Has anyone tried this approach before? If it didn't work - why not, and what did?
Asked by Dominic Demierre (1 rep)
Nov 7, 2024, 05:28 PM