The following message is a courtesy copy of an article that has been posted to bit.listserv.ibm-main,alt.folklore.computers as well.
[EMAIL PROTECTED] (Gerhard Postpischil) writes: > Back in the seventies I was in charge of the systems group at a > service bureau. One of our customers was from a local university, > running an APL application that tracked students vs. classes, and a > few other things. It was gold mine - whenever it ran, the CPU went > 100% busy and stayed that way for a long time. The same thing written > in another language might have cost one or two percent as much. recent post mentioning world-wide hone system http://www.garlic.com/~lynn/2007.html#30 V2X2 vs. Shark (SnapShot v. FlashCopy) http://www.garlic.com/~lynn/2007.html#31 V2X2 vs. Shark (SnapShot v. FlashCopy) http://www.garlic.com/~lynn/2007.html#46 How many 36-bit Unix ports in the old days? http://www.garlic.com/~lynn/2007b.html#51 Special characters in passwords was Re: RACF - Password rules http://www.garlic.com/~lynn/2007d.html#39 old tapes http://www.garlic.com/~lynn/2007e.html#38 FBA rant http://www.garlic.com/~lynn/2007e.html#41 IBM S/360 series operating systems history http://www.garlic.com/~lynn/2007f.html#12 FBA rant http://www.garlic.com/~lynn/2007f.html#20 Historical curiosity question http://www.garlic.com/~lynn/2007g.html#31 Wylbur and Paging http://www.garlic.com/~lynn/2007i.html#34 Internal DASD Pathing http://www.garlic.com/~lynn/2007i.html#77 Sizing CPU http://www.garlic.com/~lynn/2007j.html#65 Help settle a job title/role debate http://www.garlic.com/~lynn/2007k.html#60 3350 failures HONE ("hands-on") started out in the US with cp67 ... sort of to allow branch office SEs to have "hands-on" with various operating sysetms (running in virtual machines). prior to 23jun69 unbundling announcement http://www.garlic.com/~lynn/subtopic.html#unbundle a lot of SEs got much of their "hands-on" experience in their customer accounts. after the unbundling announcement, SE time was being charged for ... and not a lot of customers were interested in paying to have SEs learn. however, the science center http://www.garlic.com/~lynn/subtopic.html#545tech in addition to doing virtual machines, cms, inventing GML (precursor to SGML, HTML, XML, etc) http://www.garlic.com/~lynn/subtopic.html#sgml and the internal networking technology http://www.garlic.com/~lynn/subnetwork.html#internalnet which was also used in bitnet and earn http://www.garlic.com/~lynn/subnetwork.html#bitnet also did a port of apl\360 to cms (cms\apl). apl\360 had its own monitor, scheduler, workspace swapping, terminal handler, etc ... all of which could be discarded in the port for cms\apl. also in moving from the 16kbyte (sometimes 32kbyte) "real" workspace sizes .... to CMS ... where the workspace size could be all of virtual memory ... the whole way that APL managed storage had to be reworked (the real storage stategy resulted in enormous page thrashing). part of cms\apl was also the ability to access system services (things like read/write files) ... something that apl didn't previously have. the combination of really large workspace sizes and the access to system services ... opened up APL for a lot of real-world problems. A lot of modeling off all kinds was done ... as well as a lot of stuff that these days are implementing with spreadsheets. One of the early "big" APL uses (at cambride) were a number of business planners from corporate hdqtrs in armonk. they forwarded a tape to cambridge with all of the most sensitive corporate customer business data ... and would do significant amount of business modeling and planning. this created an interested "security" scenario for the service at cambridge since there were a lot of non-employees using the system from various educational institutions in the cambridge area. one instance is this slightly related DNS trivia topic drift ... more than a decade before DNS http://www.garlic.com/~lynn/2007k.html#33 Even worse than Unix Before long there were a significant number of CMS\APL applications written that supported sales & marketing and deployed on the HONE system ... effectively taking over its whole use for sales & marketing (and eliminating the original "hands-on" use for SEs). Before long, sales couldn't even submit customer orders that hadn't been processed by some CMS\APL application. HONE transitioned from cp67 to vm370-based platform and from cms\apl to apl\cms (enhancements done by the palo alto science center ... including the 370/145 apl microcode assist) ... and clones of (US) HONE system were sprouting up all over the world (some of the early ones i even got to handle ... like when EMEA hdqtrs moved from the US to Paris). lots of other posts mentioning HONE and/or APL http://www.garlic.com/~lynn/subtopic.html#hone in the mid-70s, the US HONE datacenters were consolidated in silicon valley. The large customer base (all US sales and marketing) drove the requirement for large disk farm ... and the heavy APL processor requirements helped contribute to it (probably) being the largest single system image installation in the world at the time ... aka large number of loosely-coupled processors with front-end that would route new logins based on availability and load. later the silicon valley HONE datacenter was replicated first in Dallas and then a 3rd replica in Boulder (with availability and fall-over across the three datacenters). the science center had also done a lot of performance tuning and modeling ... including much of the stuff leading up to capacity planning. a sophisticated science center performance model written in APL was made available on the HONE system as the "performance predictor" ... where sales & marketing could provide workload and configuration characterizations of a customer installation and ask "what-if" questions about what happen when workload &/or configuration changes were made. when i was getting ready to release my resource manager ... besides getting selected to be guinea pig for charging for kernel software ... I had developed a bunch of automated benchmarking processes. there was something like a 1000 benchmarks defined that covered a broad range of configurations, workload, and performance tuning parameters. as each benchmark was run ... the results were fed into a specially modified version of the performance predictor. then the performance predictor was programmed to start selecting various tuning, configuration and workload specifications ... for something like another 1000 benchmarks (predicting what should happen in each benchmark and then verifying the results). the 2000 benchmarks took something like 3mnths elapsed time to run ... http://www.garlic.com/~lynn/subtopic.html#bench leading up to the release of the resource manager. httl://www.garlic.com/~lynn/subtopic.html#fairshare http://www.garlic.com/~lynn/subtopic.html#wsclock ---------------------------------------------------------------------- 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