The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well.
shmuel+ibm-m...@patriot.net (Shmuel Metz , Seymour J.) writes: > Yes, and you could add more at much more attractive prices than external > ECKD DASD. Admittedly the I/O bandwidth was much lower, but for an entry > level system that's not an issue. the original definition of RAID (from 1987) was: Redundant array of inexpensive disks (although somebody that I worked with a generation earlier at san jose got a patent on the technology a generation earlier) http://en.wikipedia.org/wiki/RAID one of the raid configurations is purely parallel transfer. with fiber channel transfer speeds ... at the time when there was struggles to get mainframe disks at escon speeds. old post about jan92 meeting wanting harrier to turn into interoperabiilty with fiber channel ... instead of SSA http://www.garlic.com/~lynn/95.html#13 what i remember of the fiber channel standards from the early 90s ... where the escon forces was creating quite a bit of contention trying to layer significantly more complex structure on top of the native FCS to handle ECKD half-duplex, synchronous type stuff (subsequently turns into FICON). part of several Harrier, SSA, SCI, and FCS activities (from the late 80s & early 90s) were taking parallel bus protocol, packetizing them and running them asynchronously over (dual simplex) serial links ... significantly improving thruput and eliminating the end-to-end synchronous latency required by half-duplex parallel bus protocols. Harrier could use high-end SCSI disks ... with dual 80mbit/sec serial copper links (dedicated serial link for each direction of data flow). As part of HA/CMP http://www.garlic.com/~lynn/subtopic.html#hacmp I benchmarked a wide variety of workloads on Harrier and standard SCSI ... using effectively identical disks ... with Harrier having significantly higher thruput ... especially as load went up. Harrier then evolves into SSA http://en.wikipedia.org/wiki/Serial_Storage_Architecture By comparison, in that time-frame, ESCON forces were fighting to emulate parallel bus synchronous operation on top of FCS (for what later becomes FICON) ... which was nearly the opposite of most of the other direction from the period. ESCON basic was emulating half-duplex parallel over 17mbyte/sec links (getting 17mbyte/sec aggregate modulo syncrhonous latency delays). About the time that ESCON finally shipped, RS/6000 also came out with SLAs. ESCON had been kicking around POK for a long time before finally getting out. In that period, one of the RS/6000 engineers took the ESCON definition, tweaked it to be about 10% faster, made it full duplex/asynchronous (i.e. 220mbit concurrent in each direction, 440mbit/sec aggregate) and used significantly less expensive optical drivers. The engineer was then convinced to not do a 800mbit version of SLA ... but to work on FCS standard instead. that engineer was then also heavily involved in cluster scaleup ... old medusa email http://www.garlic.com/~lynn/lhwemail.html#medusa -- 42yrs virtualization experience (since Jan68), online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html