Jochen - hard to say.
The argument for putting everything on one massive aggregate is so that
you have the performance you need with the existing disk capacity (vs
having multiple aggregates\volumes each with their own parity disks).
However, planning for failure, it is possible but very unlikely for an
aggregate to go offline due to some sort of failure. That being said,
you might lose the filer itself during a disaster.
The documentation states that everything should be in a separate volume,
and sites Microsoft as stating this. Undoubtedly, MS states this for DR
reasons, while NetApp states it for issues with snapshots (and with
tradvols, for DR reasons).
After giving this much thought, my recommendation would be one huge
aggregate with multiple volumes - it gives you the performance you need
without wasting disks. If you really need to worry about DR, then
RAID-DP and SnapMirror are the right solutions (to prevent failure and
to deal with failure when it happens, respectively).
Glenn
-----Original Message-----
From: owner-toasters(a)mathworks.com [mailto:owner-toasters@mathworks.com]
On Behalf Of Willeke, Jochen
Sent: Thursday, March 30, 2006 1:46 AM
To: toasters(a)mathworks.com
Subject: RE: snapinfo -- exchange on netapp
Hi everybody,
thanks a lot for your help. I think i will have a look at the
snapmanager documents.
To the thing with the two aggregates...I plan it this way because i want
to have the logs and the DB's on different harddrives. Do you thing the
gear i get out of putting all into one aggregate is worth having less
security?
Best Regards
Jochen
-----Original Message-----
From: Glenn Walker [mailto:ggwalker@mindspring.com]
Sent: Wednesday, March 29, 2006 11:39 PM
To: Glenn Dekhayser; Sto RageC ; Willeke, Jochen
Cc: toasters(a)mathworks.com
Subject: RE: snapinfo -- exchange on netapp
The SME admin guide walks you through sizing the SnapInfo LUN.
To paraphrase (based on my memory):
number of transaction logs generated between backups * 5MB * number of
backups you want to keep online
So if you do a backup every hour and want to keep 7 days worth and each
backup has 100 logs:
100 * 5 * 7 = 3500MB
Obviously you want the LUN to be just a wee bit bigger than 'absolutely
full'.
Exchange doesn't need it - SnapManager for Exchange needs it so that it
can take your logs from the 'live' repository and move them during every
backup - it is only used for recovery (which would inclue log shipping
for DR).
Glenn (the original one?)
-----Original Message-----
From: owner-toasters(a)mathworks.com on behalf of Glenn Dekhayser
Sent: Wed 3/29/2006 3:18 PM
To: Sto RageC ; Willeke, Jochen
Cc: toasters(a)mathworks.com
Subject: RE: snapinfo -- exchange on netapp
Be VERY careful here. If your SnapInfo Lun isn't big enough, SME
backups WILL fail (experience talking here). It needs to be big enough
to house (potentially) the entire transaction log, plus some.
Glenn (the other one again).
________________________________
From: owner-toasters(a)mathworks.com [mailto:owner-toasters@mathworks.com]
On Behalf Of Sto Rage(c)
Sent: Wednesday, March 29, 2006 1:15 PM
To: Willeke, Jochen
Cc: toasters(a)mathworks.com
Subject: Re: snapinfo -- exchange on netapp
Hi Jochen,
Snapinfo-lun is needed for SnapManager for Exchange(SME). This is
where SME stores its information for recovery. Its basically used when
running "up to the minute recovery", where it copied all the
transcaction logs before it truncated them while creating the snapshot
backups. So when you do an "up to the minute" recovery from older
snapshots, it uses this location to bring in the older transaction logs
to be replayed.
Why do you want to configure 2 separate aggregates? did you mean
volumes? I'd rather have a very large aggregate with as many disks I can
get hold of instead of 2 smaller aggregates. It definetly improves
performance.
The way we have it setup in our environemnt is 1 large aggregate with 3
volumes per exchange host..
- 1 vol for the storage group with luns for each store
- 1 vol for log lun
- 1 vol for snap info lun
We also have other CIFS shares on volumes/qtrees within the same
aggregate.
Hope this helps
-G
On 3/29/06, Willeke, Jochen <Jochen.Willeke(a)wincor-nixdorf.com> wrote:
Hi everybody,
we are setting up an exchange 2003 system on a filer. Right now
i am
about to size the system. I often read that i need to have a
snapinfo-lun. But i do not understand where i need this lun for.
Exchange does not need it, does it?
What i want to configure are two aggregates. One containig
qtrees with
luns for the databasefiles and one aggregate with qtrees and
luns for
the logfiles. And if really needed i would put the snapinfo-lun
into the
logfile aggregate.
Perhaps somebody has already set up an exchange server on netapp
storage
and can help me.
Thanks a lot in advance
Jochen