IBM’s upscale Ultrastar 9LP is positioned directly against other enterprise-class drives such as Seagate’s Barracuda 9LP and Quantum’s Atlas III. The lower-end Ultrastar 9ES posted some impressive figures in ZD’s WinBench (along with some anomalous ThreadMark results) in addition to providing quiet, cool operation. The 9LP promises more performance and reliability, albeit at a significantly higher price than the 9ES.

The Ultrastar 9LP stores its 9.1 gigs of data across 5 1.8 gigabyte platters - standard for today’s low-profile, 9 gig SCSI drives. Seek time is rated at an impressively low 6.5 milliseconds. 7200rpm spindle speed and a 1 meg buffer round out the specs. The unit is backed by a five year warranty. The 9LP is available in both Ultra Wide and Ultra2 SCSI versions. For this review, we’ve taken a look at the Ultra Wide DGHS09U.

Like all the other SCSI drives the Storage Review has used, the Ultrastar 9LP arrived in a nondescript plain cardboard box. There literally was no documentation at all accompanying the drive. A quick check at IBM’s website provided the necessary termination and ID-setting information.

Before I test any SCSI drive, I use Adaptec’s ThreadMark to check the unit’s caching settings, ensuring that both the read and write boxes are checked. The 9LP, like a couple other drives, arrived with write-caching set off. After enabling write-caching, I moved on with some WinBench 98 tests. As I was about to move onwards to some ThreadMark trials, I noticed that write-caching was still disabled despite my earlier precautions. For some reason, it turns out that the ThreadMark caching changes wouldn’t take. I installed Adaptec’s EZ-SCSI 5.0, hoping that the program’s SCSI Explorer feature would enact a more permanent change on the setting. This time, write-caching stayed on. I restarted all tests.

Aside from the lack of documentation and the stubborn cache settings, installation of the drive went smoothly. ZDBop's Winbench 98 along with Adaptec's Threadmark 2.0 were both run on the unit in Windows 95 OSR 2.1 and Windows NT Workstation 4.0. The drive was partitioned into a single volume of maximum size. The average of 5 trials are presented below.

The Ultrastar 9LP turned out outstanding figures under WinBench 98, posting the highest figures we’ve yet seen for a 7200rpm drive in both Windows 95 and Windows NT, edging out Seagate’s Barracuda 9LP by a margin of 2-6%. The Barracuda tenaciously held onto its lead in ThreadMark by a miniscule 1-2% gap. So, performance and price for both drives are similar. But-

For a high-performance SCSI drive, the Ultrastar 9LP runs remarkably quietly and cool. Don’t get me wrong, the drive is still louder than the typical ATA (and thus the Ultrastar 9ES) drive. Even so, its significantly quiter than the churning Barracuda. I’d characterize the seek of Seagate drives, both Barracudas and Cheetahs, as a low, deep, full rumble. The Ultrastar’s seek noise is a much thinner, more hollow sound, yielding a much less obtrusive noise footprint. The drive ran moderately warm to the touch outside a drive cooler; with active fan cooling, it felt no hotter than any ATA drive. Two definite plusses for those looking for high performance without the drawbacks.

The Ultrastar is a clear choice over the Barracuda. Identical if not better performance combined with lower acoustics and heat at the same price. A tougher decision actually arises when we take a look at the Ultrastar 9LP vs the Quantum Viking II. The Viking is less pricey and offers performance approaching that of the Ultrastar. In addition the lower performance, though, you’ll gain a bit more noise. In my opinion it’s a pretty "linear" relationship: You get what you pay for with the increase in price- the 9LP is not a better nor a worse buy than the Viking II. If price is no object though, but heat/noise considerations are, there is no better drive you can choose than IBM’s Ultrastar 9LP.