2026-10-05 21:29 UTC
Once again looking at contract wording for things and seeing "99.9% percentage of time that the backups are performed on time and free from errors and omissions"
It seems fine until you go to actually apply it. What time frame? A day, a month, a year?
Let's say a month, and our backup policy includes 10 resources backed up daily. A single failed backup drops us to 99.67%
That's fine, we just run a bunch of reliable backups at the end of the month to catch up?
Some of our backups are continuous point in time recovery. Do these count as 1 backup or 20 billion? If the continuous backup needs to restart is that a failure?
Replies (1)
-
@napili@hachyderm.io 2026-10-05 21:37
@xssfox@cloudisland.nz the difficulty defining a meaningful SLO! Does 99.9% mean one failure at random out of a thousand? Or that backups are offline for 44 minutes a month? https://en.wikipedia.org/wiki/High_availability