Sparkling results have bought a confident tone to Western Digital. Like Seagate and in contrast to struggling Maxtor and loss-making Hitachi GST it has orchestrated the various HDD manufacturing and marketing elements to produce sweet financial music.
Western Dig hasn't been in the enterprise drive space for a long time. Now, with a more reliable version of its SATA drives it is in there and has appointed Hubbert Smith, ex-Intel, to be its director of enterprise marketing. Hubbert working base is Lake Forest, south of Los Angeles. His home is near Provo in Utah and the man accumulates air miles faster than a disk drive accumulates bad blocks with his weekly commute.
What he's flown into the UK to tell us is that Western Dig has a great RAID Edition 2 400GB SATA drive: "Non one else has this reliability at this capacity. The MTBF is 1.2 million hours plus." He epects the annual failure rate (AFR) to be below 0.7 percent: "What we do know is the previous generation RAID Edition is holding below 0.7 percent. And we do know the 400G RE2 reliability demonstration test exceeded 1.2 million hours in a 100 percent duty cycle, 100 percent power on hours calculation. So we have good confidence the 400 RE2 will be significantly below 0.7 percent."
Hubbert makes much of the AFR and of Western Dig's publication of it: "It's real world data. No other vendor publishes it." The implication is that there is no good way to calculate how reliable their products are. Of course Western Dig needs to establish the reliability of its enterprise SATA drives because SATA drives are regarded as being not reliable enough for enterprise use, for a 24x7 100 percent hours-on duty cycle.
He says: "The Caviar RAID Edition 2 drive is a 4-platter design leveraged off desktop drives. There have been design improvements, an extended test process to reduce 'infant mortality' and improve the product (so as) to lower the failure rate. Some Raptor mechanisms are re-used."
The extended test process is all part of a Six Sigma process, a manufacturing quality discipline which Seagate also uses.
We pressed Hubbert about SAS: "Surely WD must produce a SAS drive?" No way Jose, was the response: "WD has no plans to produce a drive with a SAS interface. We're providing the best SATA drive for SAS environments (because) we have established enterprise SATA."
The thing is that a SAS drive interface doesn't provide any drive I/O benefits and would entail WD in developing a SAS protocol stack: "To build SAS means you have to have a SAS stack. It's not easy."
SATA 1.5 runs at around 150MB/sec per channel. Drives dump data into a channel at c102MB/sec for 15,000rpm, 70-74MB/sec for 10K, and c65MB/sec for 7200rpm. These are industry-average figures.
In other words a SAS drive interface wouldn't increase the rate at which drives pour data into the link. Until a drive maxes out the link between it and the drive array interface there is no need to change the link.
Hubbert reckons SAS drives will dominate the performance market and SATA the capacity one.
What about faster SATA drives from WD? "We have a 10K drive Raptor SATA drive that will compete with SAS. There is no comment on plans to have a 15K SATA drive." Clearly though, a 15k SATA drive would enable WD to compete even more strongly in the enterprise space.
So Seagate can relax and WD's channel parts give up pressing for faster ATA drives because WD isn't going to produce a faster drive: "That's to be determined. I'm not in a position to pre-announce a product."
You get the impression that he wishes he were. Like a politician, Hubbert comes across as 'a pretty straight kind of guy', one who wraps a warm cocoon of words around you and assuages your concerns without committing himself too much about future plans - but there is lee way in the 'too much': "It's the disk drive business. We will do faster. We will do larger. It's only a question of when."