ATA vs SCSI: The last word?


— 12:00 AM on December 13, 2001

The intent of StorageReview's recent article was to show off their new test bed: test bed three. Like a fine wine though, the years have have only matured the SR staff and it shows in their testing methodology, their choice of tools, and even their choice of platform.

The shift to test bed three also called for a rematch of today's modern drives. The results of the past on the old test bed simply won't be relevant in comparison against any drive benchmarked in the new test bed. As such, the unveiling of test bed three called for a modern hard drive round up the likes of which, I dare say, have never been attempted by any sane person.

The breadth of the article easily kills two birds with one stone. Those of you interested in a new hard drive simply couldn't find a better guide to base your descision upon. At the same time, though, the article manages to exult the pros and cons to the age old storage interface war—a pro and con list that may prove to be a sour pill to swallow for many.

Workstation and home use are under siege by ATA drives. Even with more subjective measurements like boot time, ATA drives are powering their way into the lead—putting some 10,000 RPM SCSI drives to shame. This isn't to say SCSI isn't the top dog still. Certain well-designed 10,000 RPM drives (Atlas III) and, of course, the Seagate 15,000 RPM drives naturally refuse to surrender. However, when factors like price-performance ratio, price per megabyte, and even environmental operation are compared to those same 10,000 and 15,000 RPM SCSI drives, ATA's showing looks even better.

Of course it would be negligent not to show that SCSI still has a few tricks up its sleeve. Like absolutely bone shattering server performance. Under conditions of wildly random reads, SCSI's seek time simply cannot be touched—all courtesy of more advanced spindle speeds. There is also the matter of reduced CPU usage, albeit the results are more indicative of the quality of the implementation of the host controllers used, Intel ICH2 and Adaptec Ultra160, and their drivers.

SR manages to debunk some myths, and I must say it's quite refreshing. As a bit of an added bonus, SR also manages to test the value of using smaller partitions to improve stroke time.

This is reading that simply should not be passed up.

 
   
Register
Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.