chris at northlandusa
Mar 1, 2012, 2:46 PM
Post #3 of 3
(312 views)
Permalink
|
|
RE: snapshot bolus after re-enabling de-dup
[In reply to]
|
|
That's expected behavior because the snaphots are holding onto the blocks that dedupe is removing. Snapshot usage will drop back down once the snapshots stale out which looks to be about one week with your output. That's why it's always best to enable deduplication before enabling snapshots on a volume.
I believe TR-3505 has a section on this.
From: toasters-bounces [at] teaparty [mailto:toasters-bounces [at] teaparty] On Behalf Of Fletcher Cocquyt
Sent: Thursday, March 01, 2012 4:29 PM
To: toasters [at] teaparty
Subject: snapshot bolus after re-enabling de-dup
We re-enabled dedup on one of our volumes yesterday and observed the 'df-sh' savings go from 20% to 37% (yay!)
but the volume usage jumped up
snap list revealed a big delta from our usual nightly snapshot delta:
> snap list vm65net
Volume vm65net
working...
%/used %/total date name
---------- ---------- ------------ --------
0% ( 0%) 0% ( 0%) Mar 01 13:47 irt-na03(1573989836)_vm65net.450 (snapmirror)
37% (37%) 27% (27%) Mar 01 00:00 nightly.0
39% ( 4%) 29% ( 2%) Feb 29 00:00 nightly.1
41% ( 5%) 32% ( 2%) Feb 28 00:00 nightly.2
42% ( 4%) 33% ( 2%) Feb 27 00:01 nightly.3
43% ( 4%) 35% ( 2%) Feb 26 00:00 nightly.4
44% ( 4%) 37% ( 2%) Feb 25 00:00 nightly.5
45% ( 4%) 39% ( 2%) Feb 24 00:00 nightly.6
Is this expected behavior?
(when I noted these observations I was asked "doesn't dedup work on the snapshots too?")
thanks
Fletcher
|
