2.5" 15K SAS2 146GB vs. 3.5" 15K SAS2 600GB

[Piano Man]

Need some sysadmin help if possible. I have database server that I'm looking to upgrade due to capacity. Currently we have 12 2.5" 15K SAS2 146GB in RAID 1+0. We are looking at getting a storage array with 12 3.5" 15K SAS2 600GB and putting it in RAID 1+0.

My general question is in regards to IOPS. Basically 146GB and 600GB at 15K are top dogs of 2.5" and 3.5".

I would assume that 2.5" would have an advantage in regard to IOPS simply because of less travel distance. Does anyone have a comparison of some drives? I'm unable to find any.

 

[VirtualLarry]

I thought that in order to spin at 15K RPM, physics dictated that the platter size had to be smaller, thus the 2.5" form-factor.

Are you saying that they make 15K 3.5" HDs? That truely use 3.5" platters?

 

[Piano Man]

I thought that in order to spin at 15K RPM, physics dictated that the platter size had to be smaller, thus the 2.5" form-factor.

Are you saying that they make 15K 3.5" HDs? That truely use 3.5" platters?

Wow, never thought about that. There are 3.5" 15K SAS HDDS that weigh in at 600GB, but I'm not sure about the platter size in those drives. I just did some quick google-fu, but didn't come up with the patter size yet on those drives.

 

[KentState]

I know they were pretty damn quick in the EMC san I worked with. Never had a good way to compare them to the smaller 2.4" 146GB drives.

 

[Pariah]

As spindle speed increased, platter diameter decreased. 3.5" 15k drives have had platters in the 2.6" range right from the first one over 10 years ago. They shrunk a little when moving to the 2.5" form factor, but that's not where the main performance gain is noticed. The main benefit of the smaller form factor is higher spindle density in servers. More drives in the same size enclosure means more performance from the same sized enclosure. The smaller form factor also means few platters can be squeezed into the case, so read/write head mass is reduced which also increases performance on a per drive basis.

I would assume that 2.5" would have an advantage in regard to IOPS simply because of less travel distance.

The 2.5" drives are slightly faster than 3.5" drives, but as I mentioned above, the biggest benefit is that you should be able to get the same number of IOPS you are getting from your 3.5" array in an enclosure less than half the size by moving to same number of 2.5" drives.

 

[Piano Man]

As spindle speed increased, platter diameter decreased. 3.5" 15k drives have had platters in the 2.6" range right from the first one over 10 years ago. They shrunk a little when moving to the 2.5" form factor, but that's not where the main performance gain is noticed. The main benefit of the smaller form factor is higher spindle density in servers. More drives in the same size enclosure means more performance from the same sized enclosure. The smaller form factor also means few platters can be squeezed into the case, so read/write head mass is reduced which also increases performance on a per drive basis.



The 2.5" drives are slightly faster than 3.5" drives, but as I mentioned above, the biggest benefit is that you should be able to get the same number of IOPS you are getting from your 3.5" array in an enclosure less than half the size by moving to same number of 2.5" drives.

Good Info. We're not rack space bound, and I don't see us being so in the forseeable future. We're just looking to get bigger capacity without sacrificing IOPS.

So if 3.5" drives have platters around 2.6" and 2.5" drives have platters a little less than that, then I would assume density to be greater on the 3.5" drive that has 600GB (4 platters) than the 146GB 2.5" drive (2 platters). So perhaps IOPS might even improve a little?