ONTAP Discussions

Raid group size for 1TB drives?

mgalloway
2,345 Views

What is the recommened RG size for 1 TB drives when you want to maximize aggr capacity over performance?

1 ACCEPTED SOLUTION

stratify_tc
2,345 Views

There are a number of considerations in this.

First, we should operate under the assumption that you are using Data OnTap 7.x.  Within 7.x SATA RAID groups are limited to 16 disks.  It is generally accepted that SATA disks are put in to 14 disk aggregates, thus this would yield an aggregate with about 8.29TB usable (with a standard 5% snap reserve).  If you go the to max, 16 disks, you will get 9.68TB usable (again with the standard 5% snap reserve).

Second, you now need to determine if you are willing to use a second (smaller) raid group to take your total aggregate size closer to the 16TB limit.  With that said the max that you will get from an aggregate using 19 data disks is going to ~15.732TB raw (or about ~13.13TB usable with 5% snap reserve).  Note, that I have switched terminology here.  I do this for good reason, in my particular environment I have never seen the benefit of trying to get out to the logical limits so I don't have the exact numbers (note that the ~ is in front of calculated numbers, the others are from one of my systems).  You could end up a RAID of 16 disks, and then one with 7 disks (or 14 and 9 depending on you RAID group size).  Either way you go you are going to end up with 19 data and 4 parity for a total of 23 disks, this is shy of the 28 or 32 total disks that you would have from two normally created RAID groups of 14 and 16 disks (respectively).

On a small scale this may night seem that daunting, but as you scale your environment up, you are using more parity disks, thus decreasing the total size of the system.  If we extrapolate the numbers out further, and say look at what two loops (168x 1TB disks) would get you,

- With traditional 14 disk aggregates:

  - 12 Aggregates (zero drives spare)

  -  99.48 usable TB @8.29TB each

- With traditional 16 disk aggregates:

  -  10 Aggregates (8 drives spare)

  -  99.68 usable TB @9.68TB each

- With 23 disk aggregates:

  -  7 Aggregates ( 7 drives spare)

  -  91.91 usable TB @13.13TB each

Ultimately the 16 disk aggregates becomes the most equitable to use.  This yields the largest number of usable TB and spare drives.

In Data OnTap 8.0 this changes dramatically, and depending on the hardware that you are running you can see up to 100TB aggregates (FAS6080).

I hope that this helps rather that confuses.

Regards,

TC

Normal disclaimers apply, as to each environment being configured differently, and so on and so forth.  This is not a definitive guide, if it was you would have gotten it from a KB on the NOW site.

Message was edited by: stratify_tc - Addressed minor spelling errors.

View solution in original post

1 REPLY 1

stratify_tc
2,346 Views

There are a number of considerations in this.

First, we should operate under the assumption that you are using Data OnTap 7.x.  Within 7.x SATA RAID groups are limited to 16 disks.  It is generally accepted that SATA disks are put in to 14 disk aggregates, thus this would yield an aggregate with about 8.29TB usable (with a standard 5% snap reserve).  If you go the to max, 16 disks, you will get 9.68TB usable (again with the standard 5% snap reserve).

Second, you now need to determine if you are willing to use a second (smaller) raid group to take your total aggregate size closer to the 16TB limit.  With that said the max that you will get from an aggregate using 19 data disks is going to ~15.732TB raw (or about ~13.13TB usable with 5% snap reserve).  Note, that I have switched terminology here.  I do this for good reason, in my particular environment I have never seen the benefit of trying to get out to the logical limits so I don't have the exact numbers (note that the ~ is in front of calculated numbers, the others are from one of my systems).  You could end up a RAID of 16 disks, and then one with 7 disks (or 14 and 9 depending on you RAID group size).  Either way you go you are going to end up with 19 data and 4 parity for a total of 23 disks, this is shy of the 28 or 32 total disks that you would have from two normally created RAID groups of 14 and 16 disks (respectively).

On a small scale this may night seem that daunting, but as you scale your environment up, you are using more parity disks, thus decreasing the total size of the system.  If we extrapolate the numbers out further, and say look at what two loops (168x 1TB disks) would get you,

- With traditional 14 disk aggregates:

  - 12 Aggregates (zero drives spare)

  -  99.48 usable TB @8.29TB each

- With traditional 16 disk aggregates:

  -  10 Aggregates (8 drives spare)

  -  99.68 usable TB @9.68TB each

- With 23 disk aggregates:

  -  7 Aggregates ( 7 drives spare)

  -  91.91 usable TB @13.13TB each

Ultimately the 16 disk aggregates becomes the most equitable to use.  This yields the largest number of usable TB and spare drives.

In Data OnTap 8.0 this changes dramatically, and depending on the hardware that you are running you can see up to 100TB aggregates (FAS6080).

I hope that this helps rather that confuses.

Regards,

TC

Normal disclaimers apply, as to each environment being configured differently, and so on and so forth.  This is not a definitive guide, if it was you would have gotten it from a KB on the NOW site.

Message was edited by: stratify_tc - Addressed minor spelling errors.

Public