Seagate, IBM & Bad Firmware

3:05 PM -- As reported in Byte and Switch's news analysis section, IBM has admitted that the Seagate drives they've packed into BladeCenter and System X servers have the same problems as the Barracuda 7200.11 I discussed here back in December. While that blog post talked specifically about 1.5 TB drives, the problem exists across the Seagate SATA line, including the ES.2 nearline drives targeted for RAID arrays, virtual tape libraries, and the lot.

The best information I have on the problem is that one or more test systems at the Seagate plant left test data in a reserved area that caused the drive event log to essentially buffer overflow when it contained 320+(x*256) entries. Should a drive be powered up when the event log is in this state, the drive will think it's permanently busy and will not respond to external commands over the SATA interface.

If youve added any storage kit in the past nine months or so with 250-GB or larger drives you may also have affected drives, even though your OEM hasn't copped to it yet. The problem exists in Barracuda 7200.11, Barracuda ES.2, and Maxtor DiamondMax 22 drives manufactured before Jan. 19, 2009. If you have any 7200.11 or ES.2 drives you really should see this Seagate document and get the utilities Seagate's created to test to see if your drives have the bad firmware. If so, you can download and install a firmware upgrade as long as your drive hasn't already frozen up.

If your drive has frozen up, Seagate claims the data is all still there and will recover it for you, assuming of course you don't have a corporate policy against sending data out.

This being America, the California law firm of Kabateck Brown Kellner LLP set up a Website "investigating a potential lawsuit against Seagate." Sounds to me like this will end up with a class action suit where the law firm makes a bundle of money and each hard disk owner will get a coupon good for $10 rebate on a new Seagate drive.Because IBM had Seagate create a custom firmware build for the drives they sell for these servers, IBM customers can't install the firmware updates Seagate is making available to those buying retail or generic OEM versions of the drives. They'll have to wait, and hope their drives don't lock up and fail to power up, until the new firmware build passes IBM's vaunted drive qualification process, which means IBM customers, and customers of other OEMs that will similarly delay the update, have to keep their servers running and fingers crossed for the time being.

This whole affair has me wondering why we let array vendors charge us a 300 percent to 500 percent premium to buy Seagate, WD, or Hitachi drives from them. I've been using arrays from Infortrend for things like backup targets and other applications that don't need much more than a reliable place to store data ever since I noticed that they frequently were the backend storage on VTLs, archiving appliances, and similar devices that came through the DeepStorage labs. Not only do I get to buy ES.2 1-TB drives for $166 at CDW (or $140 at NewEgg) instead of the $608 EMC would charge me for one at the New York state contract price, I also get to upgrade from 500-GB to 1-TB drives without voiding my service agreement or warranty.

— Howard Marks is chief scientist at Networks Are Our Lives Inc., a Hoboken, N.J.-based consultancy where he's been beating storage network systems into submission and writing about it in computer magazines since 1987. He currently writes for InformationWeek, which is published by the same company as Byte and Switch.