Reviews Leaderboard Database Reference Search StorageReview Discussion Reliability Survey Search About Contents

Maxtor DiamondMax Plus 45 PayPal Donations

Maxtor DiamondMax Plus 45 54610H6
  September 11, 2000 Author: Eugene Ra  

IOMeter Performance

[an error occurred while processing the directive]
Despite its improved access time, the DiamondMax Plus 45 runs into trouble when compared to the best from the competition (the IBM Deskstar 75 GXP and the Quantum Fireball Plus LM). In the IOMeter Workstation Index (a normalized average of Light, Medium and Heavy loads), the Maxtor lags category-leading Quantum's offering by a significant 15%. The Plus 45's score of 142.82 actually trails that of its predecessor, the DiamondMax Plus 40, by about 1%.

What has happened here? A closer look at the individual IOMeter load scores yields an answer. By nature of its superior access time, the Plus 45 posts better scores than the Plus 40 in Linear, Very Light, and Light situations. By the time the I/O depth reaches Medium or Heavy, however, the Plus 40 leads. As a result, we've got to conclude that firmware/caching algorithms that aren't conducive to top-flight IOMeter scores are inhibiting the Plus 45's superior mechanics.

Did Maxtor's firmware tweaks that yielded improvements in WinBench 99 hurt its IOMeter scores? That may very well be the case. Why would a manufacturer do so? After all, readers know that we weigh IOMeter scores more heavily than WinBench 99 results. We should take a moment to note that even SCSI heavyweights Seagate and Quantum believe that WinBench 99 is a much more accurate measure of workstation disk performance than our IOMeter Workstation access pattern. Why? According to the engineers, while workstation disk accesses -are- predominately random in nature, these random seeks tend to occur over a tight area. In other words, the actuator will access data within a relatively tight cluster of tracks for many seeks before encountering a request requiring a long stroke. After such a movement, requested data again continues to be in a confined area. While IOMeter can simulate the random/sequential balance well, it can't in its current incarnation simulate this "localized" activity.

Again, according to manufacturer engineers, it is this localized activity that elevates the ineffable (at least at the review level) buffer/caching/firmware algorithms and strategies to the forefront when it comes to workstation performance. Thus, manufacturers such as Seagate and Quantum fully stand by WB99 results that show the Cuda ATA II and Fireball Plus LM substantially outperforming the SCSI Barracuda 18XL and Atlas V. These manufacturers do agree that IOMeter is the best measure in database and file server situations, where accesses are completely random and not localized in nature.

So, according to Seagate and Quantum, the Maxtor DiamondMax Plus 45 is one hot drive, despite its mediocre IOMeter showing .



Copyright © 1998-2005, Inc. All rights reserved.
Write: Webmaster