In terms of speeds and feeds, Symmetrix DMX is clearly an improvement over its predecessor. By moving from mirroring (RAID 0) to parity RAID, DMX has slightly more usable capacity than a Symm 8000 (37TB vs. 34TB), but uses fewer disk drives (288 vs. 384). Connectivity is up from 32 ports on the 8830 to 96 front-end ports on the DMX 2000. Last but not least, cache bandwidth improved dramatically. By switching from a bus to so-called matrix architecture, Symmetrix DMX got a 40X boost in aggregate cache bandwidth, up from 1.6GB/s to 64GB/s --compared to 15.9GB/s on Hitachi's 9980V array.
Or did it? EMC's claims to 64GB/s aggregate cache bandwidth (128 direct connections through cache at 500MB/s) is what Arun Taneja, an independent analyst in Hopkinton, MA, calls "a marketing spec number." "The more important number is 16GB/s"--the performance that a fully loaded DMX 2000, with its eight caches, can actually sustain (each with four regions, accessed simultaneously at 500MB/s),
Requires Free Membership to View
When you register for SearchStorage.com, you’ll also receive targeted emails from my team of award-winning editorial writers. Our goal is to keep you informed on the hottest topics, the latest news and the biggest challenges you face as a storage professional today.
Rich Castagna, Editorial Director
Taneja says. By way of comparison, HDS's 9980V has an effective internal cache bandwidth of 6.4GB/s.
On other fronts, DMX remains vulnerable to the HDS 9900 in terms of capacity, with only 37TBs, compared to the 9980V's 128TB. DMX is also missing a mirrored cache, although it does support a new error mitigation algorithm called Triple Modular Redundancy with Majority Voting. Only time will tell if this will satisfy users. "I'd hate to have a Florida situation," says Phil Townsend, HDS senior director of product marketing. "What happens then? A recount?"
This was first published in March 2003