Is it best to assign NetApp 4246 disks to both controllers or just one

In short, it depends, assigned to one controller will give you more disks (if creating only one aggregate versus 2).  Creating 2 aggregates, one on each controller gives you better load handling (although less spindles) on the controllers.

That depends on how many disks you have....

You will need at least 3 plus a 1 per controller.....

so how many disks do you have, and how do you want to setup this filer, for performance or storage space?

We have just gone through this process, we have a fully populated 4246, 20 HDDs and 4 SSDs.

We went for maximum performance and storage, so we create Flash Pool with SSDs, and then allocated disks all to a single controller, but had to allocate the minimum to Controller B, for HA!

If we had split the disks between the controllers, we would have had less storage, but best for performance between the two controllers.

Our biggest issue was the 4 SSDs, which can only be assigned to one controller and aggregate.

So don't forget your spares!

How many disks do you have?

Also would be a good idea to identify which version of Ontap..

The version of Ontap on the FAS 2520 is 8.2.2RC1. The FAZS 2220 at the other site is running Ontap 8.1.4P1, which I understand I need to update to the same as the FAS 2520 in order to use SnapProtect between the two.

The new FAS 2520 has 2 controllers, 12 - SAS 10K 900 GB drives (6 data, 1 spare, 1 parity, and 1 dparity). The DS 4246 has 24 - BSAS 7.2K 2 TB disks in it.

I want to set the filer up to have as much storage space as I can but also to have the performance needed to serve user files via CIFS shares, run about 5 - 7 Hyper-V VMs on it, and also to replicate backups from a remote site to it while replicating the local backups to the remote site using SnapProtect. So I'm not sure what the best approach is.

Also, I did notice the BSAS drives after I received the equipment. They were supposed to be sending me the exact setup I have at the other site, which is all SAS 10K drives. Is this going to make a noticeable impact on performance with the setup I'm looking at?

I would do it as follows - assuming 7 mode

Controller 1 - 1 aggregate of 12 900Gb drives (if you have root vol for controller2 you can move it)
controller 2 - 1 aggregate of 20 disks (max for 7.2 drives) - move over root aggr from controller 1 if on there and that will free up additional 3 or 4 900GB drives to use (fast drives aren't cheap either).  By putting all drives in same aggregate (raid group) you have the most spindles and performance.

I would update to 8.2.2 now that it was been released.

Maximum Storage would be to assign all to one controller. (and leave four disks on the other controller for HA failure).

I did consult with the Storage Solutions Architect who sold us the system and found out that his plan was all 12 900 GB SAS drives would be owned by one controller with a single aggregate for running the Hyper-V VMs and the other controller would own the 2 TB SATA drives. So it sounds like what you guys suggests is the same as what he was thinking.

So just to be sure I understand, I would want to assign all 12 SAS drives to controller one and would also need to store the root vol for that controller on there. All SATA drives then would be assigned to controller 2 with the root vol being located on that aggregate. I'm assuming the root vol needs to be stored on the aggregate that correlates with each controller? Or can I store them both on the SATA aggregate owned by controller 2 to preserve as much space as possible on the SAS aggregate?

Also, should I have 2 spares for each controller? Or is one per controller enough?

And yes, I will look at upgrading to 8.2.2..

sounds about right.  root volume for controller 2 would go on the aggr on the 4246 (so that we don't waste another 3 drives 900GB)

Each root volume needs it's on aggregate and with 7 mode you can extend that aggregate to include other volumes which is why you can't have two root volumes in the same aggr.

For the SATA you can have 2 spares (since you can only have 20 drive I think for SATA raid group).  For SAS one spare drive usually does the job since rebuilds are fast.  Another option is for the SATA shelf is to use 19 drives, 1 spare and 4 SSD drives for flash pool attached to this aggregate.  With 8.1 you couldn't attach the flash pool to a volume with root aggr but with the newer ones you can..just a thought

You still need to use SPARES for the Flash Pool as well!

So then does this sound right (just want to make sure I'm clear, this NetApp stuff is very new to me)?

All 12 SAS drives assigned to controller 1 with RAIDDP and a single aggregate that has a 180 GB root vol for controller 1, which would have 1 parity drive and 1 dparity drive as well as 1 spare drive, and then all 24 SATA drives assigned to controller 2 with RAIDDP and a single aggregate that has a 180 GB root vol for controller 2, which would have 1 parity drive and 1 dparity drive as well as 1 spare. And then from there I can create whatever volumes I need for LUNS or Qtrees and Shares?

Will do, thanks for all the expert advice.