ontap 7mode vs c-mode on FAS 2520 disk size

[atlasship]

Hi,

we just bought a FAS 2520 with 12 x 4TB setup.

Vendor suggested we downgraded to 7-mode to better utilize the disk space, so right now we have an aggregate with 19TB and another with 3TB.

NetApp Field technician was here to do a motherboard replace and strongly recommend us to go back to cluster mode, because he told me we can utilize ADP and have smaller root aggregate and claim back those disk space.

My question is, what would be the best way to go? Should we go back to Cluster mode? there's no data on the netapp, now is the best time to make the change if we want to.

Thanks!

[AlexDawson · Accepted solution]

Go to cluster mode. We have recently released ONTAP 8.2.5 for 7-Mode, and have no plans for future releases of 7-Mode ONTAP after that. With ADP you do not have to dedicate disks to boot the second controller, so you get better usable capacity. The partner who recommended you go 7 Mode for capacity needs to take better advantage of the training and tools we have available. This is from a modelling tool that we have available to partners -

7DOT - N1 = 8 drives (2 parity) = 19TB usable, 1 spare

7DOT - N2 = 3 drives (2 parity) = 3TB usable, 0 spares

7DOT total = 22TB usable

CDOT - N1 root ADP (3 slices) = 387GB

CDOT - N2 root ADP (3 slices) = 387GB

CDOT - N1 data aggr (11 data slices) = 28TB usable

Hope this helps!

[atlasship]

Thank you so much for the reply!

Yes, I did some research and it's good to get another confirmation from you.

My further question would be.

1) I think with this setup, it'd be active/passive with 2 controller, would there be a huge impact to performance? 

2) When the time comes that we add more shelves to the unit, do we need to do a major overhaul to change it back to active/active?

Thanks!

- Atlas

[AlexDawson]

1. Best practice is to keep each controller below 50% utilisation to prevent performance impact if failover is required. With an active-passive controller config, you can run the active controller to higher CPU utilisation. So effectively it's pretty similar.
2. You can add the additional disks to create another aggregate on node 2, and from that point, the system is active-active (you will probably need to add LIFs to the previously dormant node, to prevent needing to serve data via the cluster interconnect)

Glad to help!