If you consider the dropping prices on used 9gb/18gb/36gb NetApp disks, there's little reason to scrimp on an additional spare drive and/or drive shelf.

Scott

Scott Fischmann Union Computer Exchange, Inc. 6233 Idylwood Lane Minneapolis, MN 55436 email: scott@unioncomputer.com phone: 952 935 7282 | fax: 952 935 5056

Subj:RE: # of Volumes. Date:8/7/2002 3:06:28 PM Central Standard Time From:<A HREF="mailto:dheiland@lucent.com">dheiland@lucent.com</A> To:<A HREF="mailto:toasters@mathworks.com">toasters@mathworks.com</A> Sent from the Internet

We usually run with one spare for our filers, regardless of the number of disks. When I recently asked for a quote from Netapp, the sales contact suggested 3 spares for a 10 shelf system and that it was risky to run with just 1, which we've never heard from them before.

I'm not sure why its risky. We've been running this way for years, with no issues. With next day shipment of replacement disks it doesn't seem to be an issue

So how many spares do people use/recommend?

Dave

-----Original Message----- From: Matt Musgrove [mailto:matt.musgrove@thales-is.com] Sent: 07 August 2002 17:11 To: devnull@adc.idt.com Cc: toasters@mathworks.com Subject: Re: # of Volumes.

If it was me I would use 2 disks as vol0 and not keep any business critical data on it, and use the remaining disks in vol1 (with 9 disks per RAID group) with 1 disk as spare. You would be very unlucky to need 2 spares! I have 1 spare for my 28 disks. Reconstruction time shouldn't be much of an issue should it?

Matt

devnull@adc.idt.com wrote:

...

I am trying to decide on the number of volumes for our F810 which has a DS14 shelf(FC-AL), with 14 36G drives.

By default Ontap puts 8 disks in a RAID group.

I guess i am trying to figure out if

• I should have 2 volumes vol0 and vol1, one with 8 disks and one with 5 disks and i have ONE spare for both the volumes (Is this safe in terms of ratio of disks used to spares ? )

• Have 2 volumes(6 + 6) and 2 spares

• Have 1 volume with 13 disks and 1 spare. (reconstruction time on failed disk would be high, and i dont know if there will be performance issues esp in calculating parity etc)

Other suggestions are welcome too.

Many Thanks,

Regards,

/dev/null