I'm setting up a new R100 and I'm hoping someone can clarify a few things for me:
1. I was told by a NetApp sales engineer to reserve 4 disks for spares. This sounded like overkill to me. Does anyone have an opinion about this?
2. After the NearStore disks are right sized, how much usable space is left?
3. Is it a bad thing to have the root volume contain data? I will be using the NearStore to store radiology images. The image files are large, static files.
- I was told by a NetApp sales engineer to reserve 4 disks for spares. This
sounded like overkill to me. Does anyone have an opinion about this?
I cannot speak from experience, but I understand the IDE disks in this R100 are replaced fairly often. You'll want many more spares than for a production Netapp filer.
- After the NearStore disks are right sized, how much usable space is left?
Can't answer that.
- Is it a bad thing to have the root volume contain data? I will be using
the NearStore to store radiology images. The image files are large, static files.
Very bad. Let's say you have a 2x36 root volume. This would leave you with 20+GB of data space. If you wrote 15GB of data in there, then replaced those files with updated radiology images, that's 30GB total (the old files would be kept in snapshot). Your volume would overflow.
The root volume filling up is not a "good thing" (tm). In our filers, we have the root or vol0 volume be 2 disks dedicated to that purpose. NO DATA. It's a waste, but it makes for a more stable filer.
Of course, if you disabled snapshots completely, you'd have nothing to worry about in the above case. But then you'd be defeating one of the best features of filers.
/Brian/
We currently have a single volume on our filer (which includes root) and store all of our data there. The thing to watch out for in this setup is that all of your root level directories are actually qtrees which are protected by a tree quota. (I would not recommend "overbooking" your space... make sure the quotas to not exceed the total space when added up.)
I assume this theory would apply to the R100 as well.
matt
Brian Long wrote:
- I was told by a NetApp sales engineer to reserve 4 disks for spares. This
sounded like overkill to me. Does anyone have an opinion about this?
I cannot speak from experience, but I understand the IDE disks in this R100 are replaced fairly often. You'll want many more spares than for a production Netapp filer.
- After the NearStore disks are right sized, how much usable space is left?
Can't answer that.
- Is it a bad thing to have the root volume contain data? I will be using
the NearStore to store radiology images. The image files are large, static files.
Very bad. Let's say you have a 2x36 root volume. This would leave you with 20+GB of data space. If you wrote 15GB of data in there, then replaced those files with updated radiology images, that's 30GB total (the old files would be kept in snapshot). Your volume would overflow.
The root volume filling up is not a "good thing" (tm). In our filers, we have the root or vol0 volume be 2 disks dedicated to that purpose. NO DATA. It's a waste, but it makes for a more stable filer.
Of course, if you disabled snapshots completely, you'd have nothing to worry about in the above case. But then you'd be defeating one of the best features of filers.
/Brian/
-- Brian Long | | | Americas IT Hosting Sys Admin | .|||. .|||. Phone: (919) 392-7363 | ..:|||||||:...:|||||||:.. Pager: (888) 651-2015 | C i s c o S y s t e m s
-- Matt Miller IT Infrastructure Team Duke University - Fuqua School of Business
Hi David,
We have 2 R100's, and store on this moment 7 TB of radiology data on it.
1) The IDE (ATA) disk are indeed not as stable as a FC-disk, but it is not dramatic. So, I agree that 4 spares are maybe to much. I have 2 spares in the R100 and one spare disk standby in the spare kit. But more important is how you build your volumes.
We build our volumes vertical, with a RAID group of 7. When one volume has more RAID groups, we use different bridges (every shelf has 3 IDE-SCSI bridge), also for performance raisons.
When you build on this way your volumes, you can loose one shelf. We have got this 2 weeks ago (one of my employees had turn off the power supplies of one shelf by accident), and everything keeps working. We have also got one problem with a bridge (this mean: 4 disks are gone), but with the right spreading, no problem. One problem is that when one disk is failed, and replaced by a spare disk, our nice design is gone. But when we update our filer (ontap upgrade), we change the disks, so that every raid group is back one nice vertical line.
2) Every disk is almost 120 GB.
3) Because the disks are so big, I use also the root volume. But not for the continue production of radiology images, but only for some static data that I need to store. I use the 'wasted' space and I don't have the risk to overflow my root volume.
Best Regards and merry Christmas to every-one,
Reinoud Reynders
UZ Leuven Belgium
----- Original Message ----- From: "Haber, David J." David.Haber@vtmednet.org To: toasters@mathworks.com Sent: Thursday, December 26, 2002 9:49 PM Subject: NearStore questions
I'm setting up a new R100 and I'm hoping someone can clarify a few things for me:
- I was told by a NetApp sales engineer to reserve 4 disks for spares.
This
sounded like overkill to me. Does anyone have an opinion about this?
- After the NearStore disks are right sized, how much usable space is
left?
- Is it a bad thing to have the root volume contain data? I will be using
the NearStore to store radiology images. The image files are large, static files.
-
Brian Long -
Haber, David J. -
Matt Miller -
Reinoud Reynders