There have been a lot of reasons not to migrate over to 2.5-inch SAS drives since the transition from 3.5-inch drives started gaining steam a few years back. I’ve heard the idea described as ‘trading a big soft bed for two smaller hard ones’ and a few years ago there might have been some validity to the statement.
But not today. A smaller form factor, slower RPM and half the capacity doesn’t equal less density, lower performance or twice the cost anymore.
First, a little history as to why the slow transition: here was the lack of density in 2.5-inch drives, limited sources, and the big $/GB issue that made it a non-starter for most IT managers. Even here at Dell we’ve held out a lot longer than most on any kind of move off the 3.5-inch drives. The solutions out there a few short years back just didn’t make sense. Changing your entire mechanical infrastructure because things might line up isn’t a position we want to be on the wrong side of. The 3.5-inch 15K drives have meant performance and capacity for so long that switching to something smaller with lower spin-rates just didn’t cut it. So what’s changed?
Capacity jumps by 2X for 2.5-Inch Drives
We recently launched 300GB in a 2.5-inch 10K SAS drive nearly a half-year ahead of the 3.5-inch 600GB 15K offering. The biggest 3.5-inch 15K drive today is 450GB. If you stack 24 of the 2.5-inch 10K 300GB SAS drives within the same space as 12 3.5-inch 15K 450GB drives you get 7.2TB of potential storage versus 5.4TB in 3.5-inch 15K. Does this mean that 2.5-inch only makes sense until the 3.5-inch 600GB 15K shows up? Nope, not by a long shot.
You see, the density play is only half the story. Performance is a big reason to go 2.5-inch 10K drives. How can performance with 2.5-inch 10K SAS be better than 3.5-inch 15K SAS? When you use twice the number of drives within the same footprint.
The end-result is up to a 60 percent improvement in IOPS (Input/Output per second) performance for the 24 X 2.5-inch 300GB 10K SAS over the 12 X 3.5-inch 450GB 15K SAS*. This is more than respectable and much higher than the typical 20 to 25 percent step function we used to get on 3.5-inch when we stepped up from the various RPM transitions from 7.2K to 10K and from 10K to 15K.
I know a lot of folks out there are thinking things like twice the power or a heavier rack or….twice the cost! Well, believe it or not, 24 of the 2.5-inch 10K 300GB SAS drives actually consume about 20 percent less power than 12 of the 3.5-inch 450GB 15K SAS drives**. Another big surprise is that, even with twice the number of drives, they still weigh about 40 percent less than the 12 drive 3.5-inch 450GB 15K SAS solution.
We’ve touched on capacity, performance, power, and weight to compare 24 X 2.5-inch 10K 300GB SAS to 12 X 3.5-inch 15K 600GB SAS within the same 2U footprint.
$/GB, The Final Consideration
I’m not dancing around the $/GB issue here…just saving it for last. The HDD OEMs have worked hard to design the 2.5-inch 10K SAS drives to where we can get them for roughly the same $/GB (within 10 percent). This means that you can double the number of drives at 2.5-inch 10K SAS you need to hit similar capacity points in 3.5-inch 15K SAS without blowing the IT budget. Future cost trends actually show the 3.5-inch 15K capacity points losing more and more traction as industry volume shifts from 3.5-inch to 2.5-inch for tier-one mission critical SAS drives. That’s the final piece of the puzzle but a very important one as we all have a budget.
Performance, capacity, power and cost. 2.5-inch 10K SAS drives have finally delivered. All this adds up to mean the form-factor challenged 3.5-inch singer is warming up for the final song of a long show.
Have you made the move to the new form factor? Do my numbers add up? I’d welcome your thoughts.
*IOPS estimates based on random IOPS with an average Queue depth of 4, 70 percent read/30 percent write, 8k transfer per command.
**~6.2 watts for each 2.5-inch 10K 300GB vs ~17 watts for each 3.5-inch 15K 450GB