Slashdot Mirror
IBM Spins Down
beggs writes "IBM and Hitachi have signed an agreement which will take IBM out of the hard drive market in three years. This press release on IBM's web site gives some details of the deal. 18,000 IBM employees and all their hard drive related patents will join about 6,000 Hitachi employees to form a new company that will be a subsidiary of Hitachi. Sad to see big blue out of the hard drive business, they have made a lot of contributions to computing." We did a story when they announced their plans back in April.
The IBM Death Star has been defeated! The rebellion has won! :D
This event reminds me of a time when the IBM AT was the hot sh*t and IBM was going around touting their wares.
At a demo, the IBM sales rep asked for questions. My friend said "How fast is your drive?" This was at a time when 60ms access time was SOTA. The IBM rep said "80ms..." My friend retorts "But the current tech is 60ms" to which the IBM rep said "See? IBM's is faster".
Doh.
Glad to see IBM's HDD go...
I'm guessing Hitachi's going to find in a few months that they got 18,000 migrant workers and dummies propped up with sticks.
Synergy is your friend
don't forget to park the heads before shutting off the lights.
Skiers and Riders -- http://www.snowjournal.com
Sad to see big blue out of the hard drive business, they have made a lot of contributions to computing. Yeah, it's really sad. I'll espacially miss the 75gxp series.
Sector and read failures are an integral part of the ATA standard and are passed via the HD controller as responses to failures. People have NO RIGHT to complain about these failures in 75gxp, the linux kernel and fs subsystems are even designed to handle these errors gracefully and not panic. Do you complain when Java <throws> an exception? No, you put some code in the catch(e){}; Instead of complaining, do something about it, ext2 and ext3 should be adjusted so that you can use,
ext2 make install --unreliableHD-12
where the use of this switch whilst compiling ext2 will automatically incorporate RAID5-on-a-drive-Reed-Solomon-type ECC in the fs module with an ability to handle a 12percent probability of sector failure per year. The fs source code will decide the Shannon's minimum ECC distance on this information and inline the appropriate strength of ECC to absorb these failures, these extra ECC blocks will be stored on different tracks because HDs have a distinct lack of spatial ECC making them vulnerable to head-scratch and cylinder-not-found errors(?).
So there, we can all use 75gxp now, if the drive's own IDE ECC can't handle read errors, then instead escalate and use the added ECC in the ext2fs subssytem or in the kernel to perform ECC. That way the paranoid among us can hedge their bets against read failures and sector not found failures. Obviously global drive malfunctions such as total drive electronics failure or total bearing failure won't be protected against. Heck WinRAR compression has this ECC feature built in, why can't a fs which is far more critical have it built in? Quit whining.
A caveman dreams of being us, the incalculable power and riches. We dream of being Q, then what?
The 60GXP and 120GXP drives are excellent. Most of the problems that people had were from the 75GXP (older than the 60GXP) drives. Even then, the rumors of problems with 75GXPs were a little over-inflated. I don't believe that there were any problems that were more significant than anyone elses. In my opinion, there were just too many l337 h4X0r5 that were accidently killing their drives and bitching about factory defects.