Link to home
Start Free TrialLog in
Avatar of snyderkv
snyderkv

asked on

Netapp Spare

EE,

I have a netapp 2040 dual controller.

Spares are global for all aggr's but dedicated to that one controller.

So if I have Raid_DP which can survive double disk failure, don't I need two spares per controller? It just seems like everyone else is using 1 spare with Raid_DP, that doesn't make sense to me or am I missing something?

Thanks
SOLUTION
Avatar of Busbar
Busbar
Flag of Egypt image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
SOLUTION
Avatar of robocat
robocat

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of snyderkv
snyderkv

ASKER

Robocat,

But what if Raid_DP has two disk failures. It won't use the other controllers spare will it?
2 disk failures on an aggr will still allow it run.  You can also get by with a single spare. If you have a shelf you can move split the vol0 hosting vol0 between the two, this will allow you to have 2 aggrs on two controllers and save at least 3 drives versus having 2 aggrs on the internal storage and 1 on shelf.
I wouldn't do that because storage is not a problem for us. The issue is simply best practice for raid_dp. If I have to run at reduced speeds while ordering a second disk spare, then that may not be acceptable but I doubt it will run that much slower if at all.

Thanks
the spare is not per controller it is per aggregate, multiple aggregates means multiple spares
I agree.  If we split the the aggregates containing vol0 for each controller on a 2000 series unit we can save quite a few drives. Typically netapp sends out the units with both vol0 on the internal drives and if you have a shelf you create a 3rd aggr.  If there is a shelf we can move one of the controller aggrs over to the shelf and use it to serve data so that we free up 3 drives on the internal unit, process is fairly easy.
Busbar, could you back that up with docs? Because I could of swore it was per controller (global) as I read it. Which means you can have 10 Aggrs and one global spare which will provide for any of the 10 Aggrs given they are the same speed and type drives.
ASKER CERTIFIED SOLUTION
Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Thanks Paul, I'll stick to the standard build sheet since it's not a huge deal either way. We only use 15k SAS because of exchange and other high IO services

Things are going great, whether is nice this time of year and I'll be cramming on SRM shortly for some of these coop sites. That will be new. Expect more posts hehe
>But what if Raid_DP has two disk failures. It won't use the other controllers spare will it?

no, it will only use the disks that it has ownership of.

>the spare is not per controller it is per aggregate, multiple aggregates means multiple spares

This in incorrect. A spare disk is by definition a disk that is NOT assigned to an aggregate.

If you have different types of drives/sizes/speeds, you do need at least one spare of each.
Thanks