Re: INFORMIX ONLINE RAID 5 and SCO UNIX

Andy Kent in the U.K. writes: >In my experience RAID should be treated very much as guilty until >proved innocent....The only thing I can say in its defence is that >RAID 5 is supposed to be less harmful than RAID 3. Hello Andy, What you write is applicable to the older, lower RAID levels and is uncannily correct. Storage Compter first developed an operational RAID 3 back in 1984, followed shortly by operational RAID 4 and RAID 5s. The detailed test results showed that in a Unix environment on a sustained basis performance averaged only 40% of that of a single spindle drive. The company felt that sacrificing 60% of the performance for simple data fault tolerance would not be acceptable to sophisticated system users. The company continued intense RAID theory and engineering work, coming out with the dramatically faster, I/O-accelerated RAID 7 architecture a few years ago. Parallel processing, large numbers of processors, multiple layers of fast controllable cache, a real time operating system with advanced special paging algorithms and techniques, and statistically-determined parity overhead minimization processing all contribute to the measured 3X-9X performance and throughput increase that systems administrators typically get. This modern RAID architecture is currently speeding up applications right now around the world, as we speak, at over 600 very well known customers. Andy, I would encourage you to verify this to your satisfaction at Storage Computer headquarters in the U.K., near Sussex University, by calling 44 372 376834 and arranging a performance demonstration. Mike Michael Willett, EE Storage Computer Corp. 11 Riverside Street Nashua, NH 03062 Tel. 603-880-3005 :::I/O-accelerated, very fast disk arrays for all SCSI systems to 1,000 GB:::