The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well.
re: http://www.garlic.com/~lynn/2007g.html#80 IBM to the PCM market so there were two somewhat different issues in the 16mbit tr vis-a-vis 10mbit enet and the SAA scenario ... one is the overall aggregate LAN thruput as per above referenced post. the other was the design point of the individual adaptor boards. The PS2 group was heavily pressuring the 6000 organization to use all of their microchannel adapters (and not do ones of their own). there was this joke that the result was going to be that 6000s would perform like PS2s. In the SAA scenario, machines mostly interacting over the network with terminal emulation to a mainframe. As a result the individual adapter cards (including 16mbit t/r adapters) had design point of relatively low per adapter thruput. In the 6000 case with 2tier and 3tier environments, high performance workstations client has full bandwidth bursty operation ... and the servers handle the aggregate bandwidth requirements of all the assembled clients. The 6000 organization had previously done their own (16bit) 4mbit t/r adatper for the PC/RT ... and the per adapter thruput of the PC/RT 4mbit t/r card was higher than the per adapter thruput of the PS2 (microchannel) 16mbit t/r card. The SAA scenario for the 16mbit t/r card was 300 machines sharing the same 16mbit LAN bandwidth (doing mostly terminal emulation) ... so the avg. per machine bandwidth was presummed to be actually quite low (far below requirements of 2tier/3tier operation and high performance workstations) other posts mentioning SAA and/or the terminal emulation focus http://www.garlic.com/~lynn/subnetwork.html#emulation The other arrows from both the SAA and PS2 groups ... was that for a lengthy period I had been doing somewhat a weekly trends and directions posting to an internal forum ... that included the quantity one prices from the sunday sjmn. These turned out to be possibly 1/3rd to 1/5th the (official?) projected prices out of the PS2 organization ... old posts with samples (as well as summary from forrester report from the period on the projected future of the mainframe): http://www.garlic.com/~lynn/2001n.html#79 a.f.c history checkup... http://www.garlic.com/~lynn/2001n.html#80 a.f.c history checkup... http://www.garlic.com/~lynn/2001n.html#81 a.f.c history checkup... http://www.garlic.com/~lynn/2001n.html#82 a.f.c history checkup... Then the head of PS2 organization hired Dataquest to do an in-depth projection of PCs in five yrs ... and the study was to include a several hr video taped round table with a dozen leading silicon valley experts. Dataquest invited me to be one of the experts. I cleared it with my management and then explained to Dataquest that the PS2 organization wouldn't be happy to see my name ... so in the video taped introduction, Dataquest managed to garble both my name and affiliation. One of the reasons that this had come up was that we were working with Dataquest on a in-depth study of projected world-wide uptake of high-speed interconnect. This was in support of HA/CMP scale-up work we were doing ... for both numerical intensive as well as commercial applications. http://www.garlic.com/~lynn/subtopic.html#hacmp old posts mentioning some of the discussions on the commercial side http://www.garlic.com/~lynn/95.html#13 http://www.garlic.com/~lynn/96.html#15 and some old email discussing various aspects of the scale-up work http://www.garlic.com/~lynn/lhwemail.html#medusa this was about the last email in the series http://www.garlic.com/~lynn/2006x.html#email920129 in this post http://www.garlic.com/~lynn/2006x.html#3 Why so little parallelism? just before the effort was transferred (and redirected away from the commercial aspects) and we were told to stop working on anything with more than four processors. and post with old email from decade earlier on some commercial characteristics http://www.garlic.com/~lynn/2007.html#1 "The Elements of Programming Style" http://www.garlic.com/~lynn/2007.html#email801006 http://www.garlic.com/~lynn/2007.html#email801016 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
