To: gmccon who wrote (1090 ) 1/18/1998 10:12:00 PM From: David A. Lethe Respond to of 3748
I want the actual study. The press release is technically misleading. In fact, I would go as far as saying it is insulting. I am in the industry, and have done design work with software, firmware, and even UNIX SCSI device drivers in the past. I want to take the high road, and give C&L the benefit of the doubt that their study was well done, and actually condidered a wide range of application/data characteristics rather than spit out feeds & speeds of bus interfaces. The marketing people probably got out of their cages ;) For example, throughput was measured as a total of the bus interfaces x interface speed. In the real-world it doesn't work that way. A 20 MB/sec "rated" fast/wide interface can pump out anywhere between 3 MB/sec to 19 MB/second, depending on block sizes, arbitration factors, drive speeds, cache subsystem, I/O types (read, write, sequential, random), O/S used, etc. It is impossible (IMPOSSIBLE) to get the full bandwidth. Now, beating EMC (and MTI) in terms of real-world performance is very easy to do, <plug coming ..> our own Polaris subsystems cost about 1/10th as much, and range from 20% to 600% faster, depending on the data. You can be over 10,000% faster when comparing SSD to EMC/MTI. Of course the features are very different, and that is why these high-end boxes have a market, and are a good investment if you need those features. I don't want to give any people on this thread the opinion that I am slamming MTI or any other company. I am just saying that this study will mean nothing to the price of MTI's stock. There a tons of studies and whitepapers that EMC or any other RAID vendor can throw at a customer to put their product in the best light. David