The disk size and aggr snap reserve questions always seem to come up (a lot). My rule of thumb (and I think NetApp sales is getting better at positioning the product this way) is that NetApp doesn't sell space... at least not at first. They sell Disaster Recovery...

Snapshots, snapmirror, clustering. THAT is what NetApp is about - protecting your data from failure...

Next, it's performance. Lastly, actual space for data.

Protecting the data is what the aggr snap reserves are for (along with snapmirror) but they do have their limits. While filesystem corruption does occasionally happen (on EVERY platform), NetApp is protecting us from it without our needing to think about it. I personally like that approach. The other part of their approach that I like is the fact that they don't hide what they are doing like some other vendors...

Just my .02...

-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com] On Behalf Of Chris Thompson Sent: Friday, February 24, 2006 7:28 PM To: toasters@mathworks.com Cc: gdekhayser@voyantinc.com; slinkywizard@integraonline.com Subject: Re: maxing out the filesystem

"Glenn Dekhayser" gdekhayser@voyantinc.com writes:

Actually, the aggregate snap reserve is by default only 5%, not 10%.

... which is nothing to do with the 10% "reserve" I was talking about, which is at a lower level. See below.

"Max" slinkywizard@integraonline.com adds:

Somewhat tangental:

Does anyone here use aggregate snap reserves ? I was under the

impression

that is was only useful for sync mirror setups... So I usually just

shut

it off to take advantage of the extra space.

Whatever NetApp say, I've never seen the point of reserving space for aggregate snapshots if you don't _use_ aggregate snapshots for anything. We don't, so

snap reserve -A [aggrname] 0 snap sched -A [aggrname] 0

However, for pedagogic purposes, I've set it back to 5% temporarily for the example below. :-)

carina> df -A main Aggregate kbytes used avail capacity main 595174120 575814948 19359172 97% main/.snapshot 31324952 0 31324952 0%

That's a total of 626499072 KB, _including_ the snapshot reserve.

carina> aggr status main -r Aggregate main (online, raid_dp) (block checksums) Plex /main/plex0 (online, normal, active) RAID group /main/plex0/rg0 (normal)

RAID Disk Device HA SHELF BAY CHAN Pool Type RPM Used (MB/blks) Phys (MB/blks) --------- ------ ------------- ---- ---- ---- ----- -------------- -------------- dparity 8b.24 8b 1 8 FC:A - FCAL 10000 68000/139264000 68552/140395088 parity 8a.27 8a 1 11 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8b.28 8b 1 12 FC:A - FCAL 10000 68000/139264000 68552/140395088 data 8a.29 8a 1 13 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8a.25 8a 1 9 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8b.22 8b 1 6 FC:A - FCAL 10000 68000/139264000 68552/140395088 data 8b.16 8b 1 0 FC:A - FCAL 10000 68000/139264000 68552/140395088 data 8a.17 8a 1 1 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8b.18 8b 1 2 FC:A - FCAL 10000 68000/139264000 68552/140395088 data 8a.19 8a 1 3 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8a.21 8a 1 5 FC:B - FCAL 10000 68000/139264000 68552/140395088 data 8a.23 8a 1 7 FC:B - FCAL 10000 68000/139264000 68552/140395088

After the right sizing of the discs to 68000 MB = 69632000 KB, taking off 20.5 MB = 20992 KB (the reserved area at the start of each disc: it's been that ever since NetApp was hatched from the cosmic egg) that's a total of 10 (data discs) x 69611008 KB (per disc) = 696110080 KB.

The space actually made available in the aggregate to flexible volumes, their snapshots, and aggregate snapshots, old Uncle Tom Cobley and all, is 90% of that, i.e. 696110080 * 0.9 = 626499072 KB.

[It would have made a better lesson if 90% of 10 data discs hadn't equalled exactly 9 data discs! Sorry about that, you'll just have to count them carefully ...]

Exactly the same (hidden) reserve applies in a traditional volume.

[Oh, and to you guys over in the "Aggregate size question" thread, stop quoting all your figures in GB and to 2+ sig figs, as it's imposible to make anything add up if you do that. Real Programmers aren't scared of 10-figure numbers :-) ]