Maybe I should have been more explicit. Each raid group should be comprised of the same size/speed disks, and the number of disks in a raid group should be close to the recommended value for the type of drive and raid type. The only hiccup with building aggregates out of different size drives is when you want to replace the smaller older drives. You will have to migrate the whole aggregate (all the volumes) before you can destroy the aggregate and pull only the drives that you want to replace.

joel

Quoting Jeff Mohler speedtoys.racing@gmail.com:

Even so, different speeds can exist anywhere in the aggregate.

The CP to disk is NOT complete until the last block is committed, no matter where it is in the aggr...slow disks scattered across 4 RGs, or if all of the slow ones are in one RG. Its a team effort.

Raid stripes are only protection zones in disk failures, they are not independent write zones where different disk speeds cause an explicit issue when things are done correctly.

Now..if you create some small additional raid group at the end of an existing aggregate, thats a different story, but not part of 'done correctly' as stated above.

:)

On Wed, Feb 2, 2011 at 3:01 PM, Joel Krajden joelk@encs.concordia.cawrote:

...

You can specify the raid group size per raid group in an aggregate. As long as the speed of disks in the different raids groups are the same you should not see a performance issue for the aggregate. You should have at least one hot spare of each drive type and configure Ontap to use the same size drive when selecting a hot spare. Ontap should be at the latest revision if possible.

Joel

On 02/02/2011 05:24 PM, Ray Van Dolson wrote:

...

Have an aggregate comprised of 300GB disks and have two new shelves with 450GB disks.

I ran across this[1] which describes some concerns when dealing with disks of mixed sizes.

It sounds like I can't specify that 450GB disks and 300GB disks be assigned to distinct raid groups (my raid size is 16), so if I add the 450GB disks to my aggregate, I'll likely only be able to use 300GB of them.

Therefore I should create a new aggregate for the 450GB disks.

Is my understanding of this correct?

Also, the raid size seems to default to 16 -- I'm used to keeping RAID-6 raid groups to 9 disks at most. Is 16 OK?

Thanks, Ray

[1] http://now.netapp.com/NOW/knowledge/docs/ontap/rel7261/html/ontap/smg/provis...

-- | Joel Krajden | Rm: EV-7105, Tel: 514 848-2424 3052 | | Core Technologies Mgr | | | Engineering & | Email: joelk@encs.concordia.ca | | Computer Science | | | Concordia University | In a circus, the clowns are supposed | | Montreal, Canada | to make you laugh, not cry. |

--

Gustatus Similis Pullus