ONTAP Hardware
ONTAP Hardware
Hi,
I have FAS2020 (2 x node) with 12 SAS disks. I will appreciate any suggestions helping me to use these 12 disks in most effective way considering usable space.
The main thing is - is it possible to use 6 disks where Data ONTAP is installed? For example, create volumes for data in these two aggregates, o just extend aggregate there Data ONTAP is located and use single aggregate (for one filer) for both Data ONTAP and data?
If I configure system leaving default values I get only 4 disks for data (6 disks - Data ONTAP, 2 disks - spares).
TIA.
BR,
Danas
Solved! See The Solution
Several options:
Four data disks per controller: pro: utilize CPU/RAM resources of both controllers, cons: usable space per controller. No hot spares.
C1: 6 disks (4D + 2P) (no hotspare)
C2: 6 disks (4D + 2P) (no hotspare)
Eight data disks on controller 1: pro: maximize space on 1 controller, use 2nd only for cluster failover. cons: CPU/RAM resources of 1 controller. Other isn't doing anything. No hot spares
C1: 10 disks (8D + 2P) 0 HS
C2: 2 disks (RAID4: 1D + 1P) 0 HS
Seven data disks. Same as previous, but with an hot spare for controller 1
C1: 10 disks (7D + 2P) 1 HS
C2: 2 disks (RAID4: 1D + 1P) 0 HS
There are probably other options. All depends on your needs.
What you've certainly noticed is, that if you're willing to work without spares (but still use RAID-DP), you end up with 8 data disks.
Mind you some of these scenario's go against NetApp best practises, but in this case, FAS2020A with 12 disks you gotta be creative.
Regards,
Niek
Hello Danas,
There are multiple ways to do this. Minimum you need 1 parity (raid 4 aggregates), 1 spare and 1 data disk on each controller. After this you have 6 disks available. You can add these disks on one node to create an aggregate as big as possible and use the other node only for cluster failover (and only use 1 CPU for all the data).
Another possibility is add 3 disks to each node. With this setup you can use performance of both node CPU's.
Regards,
Eelco
Several options:
Four data disks per controller: pro: utilize CPU/RAM resources of both controllers, cons: usable space per controller. No hot spares.
C1: 6 disks (4D + 2P) (no hotspare)
C2: 6 disks (4D + 2P) (no hotspare)
Eight data disks on controller 1: pro: maximize space on 1 controller, use 2nd only for cluster failover. cons: CPU/RAM resources of 1 controller. Other isn't doing anything. No hot spares
C1: 10 disks (8D + 2P) 0 HS
C2: 2 disks (RAID4: 1D + 1P) 0 HS
Seven data disks. Same as previous, but with an hot spare for controller 1
C1: 10 disks (7D + 2P) 1 HS
C2: 2 disks (RAID4: 1D + 1P) 0 HS
There are probably other options. All depends on your needs.
What you've certainly noticed is, that if you're willing to work without spares (but still use RAID-DP), you end up with 8 data disks.
Mind you some of these scenario's go against NetApp best practises, but in this case, FAS2020A with 12 disks you gotta be creative.
Regards,
Niek
Does your answer imply, that I can use the same aggregate for for Data ONTAP OS and customer's data?
Thanks.
BR,
Danas
Yes.
On small FAS systems there's no point having a seperate root aggregate. It simply costs too much disks.
Regards,
Niek
OK, but how about the performance impact? Data ONTAP is not an idle OS - it constantly works calculating RAID parities etc., so IMHO one aggregate with Data ONTAP and data could have degraded performance.
BR,
Danas
Performance impact is very minimal.
Calculating raid parities is done by the CPU, not by the aggregate.
Niek
Thanks, Niek and Eelco.
BR,
Danas