20 January 2009

Correction: Snapshot Overflow

In snapshot potshot, I wrote about the "danger" of snapshot overflow. I made it sound like you can lose production space to snapshot.

Let's say you have a 100-GB qtree with 20 GB allocated to snapshot and you already used up 14 of that 20 GB. You have 85 GB of used space. You delete 10 GB hoping to see 75 GB. However, 6 GB would go into snapshot and 4 will come out of production space, yes? Or does that mean you lose 4 GB of production space? You expected 85 - 10 = 75 but instead now you now see 85 used - 6 that went into snapshot + 4 that overflowed from snapshot + 4 that could not go into snapshot so that it really stayed, meaning now you now have 85 + 8 = 93 GB !!! Luckily, a NAS expert today pointed out to me that my thinking is wrong. The 4 GB that overflowed did leave the system and simply sort of came back. So, yes, you deleted 10 GB but only see 4 GB of additional free space but that's it. You don't really lose any production space to snapshot. The 6 GB will be noticeable whenever the snapshot schedule rolls off. There, never take anything you read on the Internet for granted unless definitely independent sources can confirm it.

No comments:

Post a Comment