SAS vs. Nearline/MDL SAS - What is the difference?

As subject I guess, looking at SAN's and most vendors offer 10k or 15k "proper" SAS drives, many also offer 7.2k MDL/Nearline SAS drives.

Does anyone have an authoritative explanation of the difference please?

Thanks.

From serverfault Hutch

• Marketing.

  7.2K drives are slower and easier to produce, and with higher error thresholds which improves yields (and capacity). However, in terms of I/O operations each discrete disk can support, the 7.2K drives are markedly less performant than their faster brethren. Therefore they get the 'Nearline' moniker, as they'll hit I/O saturation much faster than an equivalent number of 10K or 15K disks. Therefore the storage producers need a way to convey, "we have faster stuff," so they went with MDL/Nearline.

  This is how they try to encourage people who need both fast and lots of storage to go for the faster drives. Those on a budget will see that you can get (for example) a 1.5TB MDL/Nearline for half the price of a 450GB 15K drive and wonder why the upcharge. Even so, 48 7.2K RPM drives will still outperform 12 15K RPM drives. It's just that the 48 7.2K RPM drives will probably have a capacity of 30TB, where the 12 15K RPMs may only have 5TB of capacity.

  Which is another way of saying...

  Go 7.2K RPM when capacity is your number one goal and performance not really a goal.
  Go with 15K RPM when performance is your number one goal, and capacity secondary.

  Hutch : Thanks, in my case I have an IOPS target and a large total capacity, some of which needs to be fairly fast, some doesn't matter. I can get a lot of capacity AND the IOPS by throwing lots of cheaper but individually slower MDL spindles at the SAN - didn't know what downside there may be as it seems too good to be true.

  Farseeker : And there was me thinking that "Nearline" was some kind of performance bonus over a normal 7.2k drive... bloody marketers...
  From sysadmin1138

• Great answer by 1138 but one other aspect of note is that midline driveS usually do not have a '24/365' duty-cycle, this is often hard to spot in the specs but it means that if you drive the disks long and hard they'll fail FAR more frequently than you'd think. As an example we had a ~200 disk array full of 1TB MDLs and saw a >20% failure rate in the first year because the array was active for 20-24 hours a day, we had to swap the lot for 10Ks.

  sysadmin1138 : Oo, good catch. Forgot the duty-cycle / life-expectancy point.

  Hutch : I'm looking at a HP drive specs PDF right now and no sign of duty cycle/MTBF - all very ambiguous as one man's backup IO could be a magnitude more than another's production IO - perhaps they see it as a question of criticality of data.

  Chopper3 : Oh goody, I'm a HP storage and server geek! Mire tan happy to help you with product selection or any other queries.

  Hutch : I have HP coming out tomorrow, it's the P4000 I'm looking at if you've any knowledge, at least 15tb usable and replicated so of course it gets a bit (make that very!) pricey if you're looking at 15k SAS.

  Chopper3 : Consider the EVA4400, I love EVAs and XPs, great kit
  From Chopper3