Estimated full takes retention policy into consideration?

[jsabater]

Hey everyone!

Quick question: the Estimated full column for a given datastore in the Dashboard: Datastore usage takes the retention policy into consideration when calculating the expected time until the disk is full?

I assume it is statistically computed given the deduplication level that has been achieved in the history of each backup, its size and growth, and probably other parametres. Would this assumption be correct, or is it a much more simple calculation?

Thanks.

 

[ggoller]

Hi!
it is a very trivial calculation, we simply take the linear regression (least squares) over the last usage data points.
More info: https://en.wikipedia.org/wiki/Linear_least_squares#Example

 

[jsabater]

By the last usage data points, do you mean the total size on disk for all backups, not taking into consideration what is a backup, what is a snapshot and the deduplication level achieved over time (and, therefore, presumed to be achieved in the future)?

Moreover, from your words I presume that it does not take into consideration the retention policy, does it?

Thanks in advance.

 

[ggoller]

Every 10 seconds, PBS retrieves the usage (with `stat`) of every datastore (disk) and writes it to the RRD. To calculate the estimate, it takes the last usage data points and calculates the linear regression on it. So it does not account for deduplication nor the retention policy.