Has IBM Solved Its iSeries Disk Problems?

by Timothy Prickett Morgan

We haven't heard much lately about the disk drive failure problems that were plaguing IBM's iSeries and AS/400 servers last year. While IBM seems to have cleared up a lot of the problems it was having with its 10K and 15K RPM disk drives, there might be some lingering problems as microcode fixes and new, modified disk drives make their way into the OS/400 installed base.

According to IBM's APAR document MA24457, the company was working on providing support for Seagate disks in the feature 6717, 6718, and 6719 disk units of iSeries and AS/400 servers. The feature 6717 drive is an 8.58 GB, 10K RPM disk; feature 6718 is a 17.54 GB, 10K RPM disk; and feature 6719 is a 36 GB, 10K RPM disk that is a year late to market and has yet to be announced, but which is nonetheless installed at some customer sites. IBM has yet to announce 15K RPM disk drives in any capacity for the iSeries line, and the 36 GB and 73 GB units running at that speed, which IBM has been trying to get to market for the past year, have apparently been redesigned and are just now getting ready for volume shipments. PTFs MF27487 for V4R5 (with a prerequisite of having PTF MF27352 installed) and MF27488 for V5R1 (with a prerequisite of having PTF MF27353 installed) update the OS/400 operating system so that Seagate drives can be used within iSeries and AS/400 servers.

This Seagate support may, at this point, however, be unnecessary. IBM's Technology Group, which manufactures the disk drives for IBM and for OEM customers, in late November issued a massive software microcode update for the 9 GB, 18 GB, and 36 GB members of the UltraStar LZX family of 10K RPM disks (these are the ones used in the IBM features outlined above). Some changes were also apparently made to earlier UltraStar LP 7200 RPM disks as well. According to the IBM OEM customers I spoke with, it is customary to have an all-encompassing disk-drive microcode patch just as a unit is coming into the market. The UltraStar 36LZX drives had two big microcode changes around the time they were announced in October 1999, and then another big change on November 29, 2001. This was apparently unusual, and it is even more unusual to have a big microcode update so long after a disk product has been put into the field.

When you sort through all of the technical gobbledygook in the engineering-change notice, the UltraStar LZX drives had microcode that was essentially insane. It would hang after overwriting vital data. It had bugs in vital pieces of the code that allowed a drive to reset after a failure. The error-recovery code and write cache microcode had bugs that caused it to hang. The code-controlling arms of the drive could cause the drive head to go off track, causing errors.

It is unclear if IBM has put this updated disk drive microcode into PTFs for iSeries and AS/400 customers. As far as I know, IBM has not issued a new DASD Fixpack since last summer, when it admitted that there were problems with iSeries and AS/400 disks. And, perhaps more significant, just because the drive microcode is fixed does not mean that other OS/400 microcode relating to disks and I/O subsystems is working exactly as intended. The word I hear is that some very large customers have recently had hundreds of IBM disks recently replaced in their iSeries and AS/400 servers. Whether they were replaced with improved IBM disks or Seagate disks is unknown.

One final note: While IBM may have provided patches to OS/400 to allow Seagate disks to work in the OS/400 server line, that certainly does not mean that you can go out and buy raw Seagate disks and drop them in your servers if you apply the patches. There would be a lot of tweaking in the Seagate drives and in the electronics surrounding them to make these drives work. This kind of work can only be done by IBM or a third party disk subsystem supplier that has intimate knowledge of the iSeries and AS/400 hardware and software stack.