ONTAP Hardware

Raid group size with FAS3140 and DS4243

blanphier
6,734 Views

I have a FAS3140 with 2 controllers and 2 DS4243 disk shelves each with 24 300GB SAS drives. What would the recommended raid group size be if the shelves are split between the controllers?

If I use 2 spares per shelf, that would leave me 22 drives for the aggregate. Using the default raid group that would give me rg0 with 16 disks and rg1 with 6 disks which is not optimal.

Would it be better to create two smaller raid groups (like a 10+2 and a 8+2), or create one large raid group?

Should I only use 1 spare per shelf to increase the second raid group size?

9 REPLIES 9

lwei
6,734 Views

I would use 1 spare per shelf, so 2 spares total.

Do you need separate aggregate for vol0? if so, that's 3 drives for aggr0. And I'd use the remaining 20 (18+2) for aggr1. If not, then I'd go with smaller and more balanced raid groups (10+2 and 9+2).

Hope that helps,

Wei

BrendonHiggins
6,734 Views

I agreed with 1 spare per shelf.  You did not say how many aggregates you wanted to create, so it is not easy to comment on raid group sizes.  I run my 1 Tb disks in RG 16 if it helps.

I have a question about your DS4243 and how you have them powered.  Are they running on single or 3 phase power?

Thanks

Bren

blanphier
6,734 Views

I was not planning on creating a separate aggregate for the root vol since I only  have 24 disks to work with.

If I use a single spare, I have 23 drives for my aggregate. Should I create one large raid group or two smaller raid groups such as a 10 +2 and a 9+2 as suggested in an earlier post?

.

I am using single phase power for the shelf.

rkaramchedu1
6,734 Views

If I use a single spare, I have 23 drives for my aggregate. Should I create one large raid group or two smaller raid groups such as a 10 +2 and a 9+2 as suggested in an earlier post?

You had mentioned that you have clustered system (2 controllers). Would you like to serve data using both of those controllers? In that case, you'll need to split the drives among the two controllers and create individual aggregates using those drives. Based on the info so far, I can take a gander at *a* simplistic config

24 drives

-2 Hot Spares (1  HS per controller)

= 22 drives

Assign 11 drives to each controller and create aggregate out of that. With the default RG being 16, you'll get 9 data drives and 2 parity drives. Depending on your future plans, this RG size may need tweaking but we don't have that info.

If you want all the data to be served only by one controller, then you can assign all the drives to that controller, keep 1 (or 2 to enable Disk Maintenance Center) for Hot Spares. That'll leave you with 22 drives. If there are no future plans on growing this aggregate, then a RG size of 11 will give you 2 raid groups (9D+2P). The default RG size of 16 will net you 2 raid groups (14D+2P, 4D+2P), which is probably not the best configuration. There are a variety of options in between - so need more info. (Larger raid group result in longer re-build times but give you better capacity)

There's another live thread here on larger RG rebuild times: http://communities.netapp.com/message/23886

HTH

blanphier
6,734 Views

I found this in the DS4243 Disk Shelf Technical FAQ document

WHAT ARE THE BEST PRACTICES FOR CONFIGURING RAID GROUPS IN FULLY LOADED CONFIGURATIONS?

For one shelf: two RAID groups with maximum size 12. (It is possible in this case that customers will configure one big RAID group of size 23–21 data and 2 parity; however, NetApp recommends two RAID groups).

·10 data + 2 parity

·9 data + 2 parity

·1 spare

For two shelves: three RAID groups with maximum size 16.

·14 data + 2 parity

·14 data + 2 parity

·13 data + 2 parity

·1 spare

WHAT ARE THE BEST PRACTICES FOR CONFIGURING RAID GROUPS IN half-LOADED CONFIGURATIONS?

For one half-loaded DS4243 shelf, NetApp recommends one RAID group.

· 9 data + 2 parity

· 1 spare

are the raid group defaults changing with the introduction of ds4243?

No, the RAID group defaults for DS4243 are the same as for DS14-class.

Thanks for the responses.

john_dewerd
6,734 Views

I hate to be contrary, but NetApp does not (necessarily) reccommend 2 RAID groups with the DS4243. It depends upon the application, and the end users tolderance in the risk/reward dept.

I confirmed this setup (simialr to yours with 2 NetApp SE's and a 3rd party NetApp consulting SE.

We have 2 FAS3160s. One has 3 DS4243 shelves fully popultated with 450SAS. It really is a crime they they call it 450, because it is really not even that large RAW. After format and aggregrate reserves, it is closer to 400GB. Anyway. We created a Radi Group for each shelf, of 20/2/2, that is, 20 data disks, 2 parity (using RAID-DP) and 2 global spares. I configured all three 5 that way. On one of the 3160's (with 3 DS4243 shelves) I made an aggregater per shelf.

Sine the 2nd (remote FAS3160 has only 2 shleves and they are going to snap vault to each other, I wanted to keep the RAID, aggregate, Volumes setup the same way. I do have a question for you  all however.

Do you feel that it is wiser (given my RAID size) to make and aggregate that spans 2 shelves or and aggreage on each shelf? Ovioulsy the one with 3 shelves would then have either 2 or 3 aggregates (IF 3 one for each shelf / RAID Group, if 2, one for 2 shelves/RAID groups and the other for the single shelf). Opinions?

jdw

srikanthmutyala
6,734 Views

Hi

We have FAS3160 with two controllers and two DS4243 loaded with SAS 600GB 24 disks. Each FAS3160 will individual control the DS4243. We are planning to configure our RAID DP with

21 Data + 2 Parity

1 Hot Spare

above all in one aggregate.

Is it OK with above configuration. We are running Oracle database about 2.5TB in size .We need more spindles for faster access. 

TIA

Sri

lwei
6,734 Views

I think it's OK. The recommendations for max raid group sizes are:

  • FC: 28 = 26+2
  • SATA: 16 = 14+2

I think SAS goes with FC.

Thanks,

Wei

bvoogelwwt
6,734 Views

Can you provide a link on the NOW site to this DS4243 Disk Shelf Technical FAQ?  I cant seem to find it.  Thanks

Public