Netapp 8200 12GB SAS & 6GB SAS 900GB Drives

[peiffer83]

Hello Netapp Community,

We have a Netapp 8200 HA running Cluster mode 9.1P6.  It has the following disks:

1 Stack of 2xDS2246 200GB 6GB SSD Drives

1 Stack of 8xDS2246 900GB 6GB SAS Drives

1 Stack of 2xDS224C 900GB 12GB SAS Drives

1 Stack of 3xDS4486 4TB 6GB SAS Drives

1 Stack of 3xDS4246 600GB 15K SAS Drives

The problem I am running into is the 900GB Shelfs.  The netapp does not differentiate between the 12GB and 6GB 900GB drives.  It will let you add a 6GB drive into a 12GB aggregate.

1 Stack of 2xDS224C 900GB 12GB SAS Drives

Controller 1 and 2 have their root volumes on these drives.

1 Aggregate with Flash Pool enabled using the rest of the 12GB drives.

2 Spares on each controller.

Controller 2 owns the 1 Stack of 8xDS2246 900GB 6GB SAS Drives.  What I would like to do is spread out the 900gb drives across the 2 controllers to spread out the load.  Right now most of the load is on controller 2.  I made a ticket with Netapp and they said it shouldnt really be a big deal but here are my concerns.

1. Lets say a 12GB 900GB drive fails.  The Netapp may just grab a 6GB SAS drive and replace it.  Or if a 6GBSAS 900GB drive fails it may replace it with a 12GB SAS drive.  I am just not sure I am overthinking it or what.

Thoughts?  In the past when you mixed 6GB and 3GB sas drives they stepped down to 3GB.  I am just wondering if I am going to notice any performance degregation if a 6GB and 12GB SAS drive gets mixed.  I will create aggregates with the same drive type, but I cannot control what the netapp does when a drive fails.  I enabled all these options in the Netapp:

HSC-NETAPP-0-01
         raid.mix.hdd.disktype.capacity                 off          only_one
HSC-NETAPP-0-01
         raid.mix.hdd.disktype.performance              off          only_one
HSC-NETAPP-0-01
         raid.mix.hdd.rpm.capacity                      off          only_one
HSC-NETAPP-0-01
         raid.mix.hdd.rpm.performance                   off          only_one
HSC-NETAPP-0-01

Thanks.

[sgrant]

Hello, the 12GB and 6GB refers to the SAS connection speed to the controllers...it is unlikely to be max'd out with HDDs only in the stack.

However, to ensure only 12GB disks are replaced by 12GB disks you can set the autoassign policy...

Configure automatic disk assignment: storage disk option modify -autoassign-policy autoassign_policy

• Use stack as the autoassign_policy to configure automatic ownership at the stack or loop level.
• Use shelf as the autoassign_policy to configure automatic ownership at the shelf level.
• Use bay as the autoassign_policy to configure automatic ownership at the bay level.

Use the stack or shelf to ensure they remain seperated.

Hope this helps.

Cheers,

Grant.

[sgrant · Accepted solution]

My bad...refers to autoassign only not the aggregate, doh!

Nope, cant think of a way either just for the aggregate. ONTAP will treat them the same since they are physically the same disks, juts different connection. But wouldnt be too worried about the performance. This is bandwidth rather than actual disk speed.