l...@garlic.com (Anne & Lynn Wheeler) writes: > oops, late 80s is 25yrs ago ... not 35yrs ... finger slip.
re: http://www.garlic.com/~lynn/2012m.html#2 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012m.html#3 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012m.html#4 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012m.html#5 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012m.html#6 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee http://www.garlic.com/~lynn/2012m.html#7 Blades versus z was Re: Turn Off Another Light - Univ. of Tennessee almost 35yrs ... almost totally unrelated ... I did CCW batch download implementation in 1980 (although the hardware I was using was from a couple years earlier, making it 35yrs) ... predating the fibre channel stuff in the late 80s ... and the zHPF/TCW last decade. thorton and cray had done cdc6600 ... cray left to form cray research and thorton left to form network systems (makers of hyperchannel). 1980 they had filled up STL (now called silicon valley lab) and were moving 300 from the IMS group to offsite building. They had been offered remote 3270 support back to mainframe in the STL datacenter ... but they found remote 3270 human factors intolerable (they were used to CMS response with local channel attach 3270 controllers). NSC HYPERChannel channel extender was suggested and I was con'ed into doing the software support. A220 channel adapter (connected to local mainframe channel) connected to A710 to telco adapter that went over channel on the local corporate campus (T3) microwave system to the main plant site and then (T1) microwave circuit to the roof of the offsite building which had the other A710 with connections to three A510s. The A510s simulated mainframe channels which is where the local channel attached 3270 controllers were connected. My support intercepted the 3270 channel program ... and converted to datastream which was written to the memory of one of the remote A510s ... where it was then directly executed. The channel program sequence was purely at the remote end w/o needing constant back&forth to the mainframe memory ... with just the necessary data transfer occuring (similar to the design point for original fibre channel from the late 80s and similar to the way that zHPF/TCW now works). similar configuration from old NSC marketing pitch http://www.garlic.com/~lynn/hyperlink.jpg This drastically cuts the chatter & latency over the interface ... as well as improving the throughput of the channel. The net was that the 300 from the IMS group didn't see any difference between small subsecond response in STL and the response at the offsite bldg. In this particular case there was a secondary side-effect. STL datacenter convention had been to distribute the local 3270s controlleres across all available channels, sharing them with disk controllers. In turns out that local 3270s controllers have rather high channel busy and were actually interferring with disk thruput. The single A220 could do all the same operations of large number of 3270 controllers with significant lower channel busy (the exorbitant 3270 controller channel busy being moved to the remote site with the A510 channel emulators). The net here (moving the local 3270s off real channels) was overall 10-15% improvement in system throughput. 3270 logon logo for IMS people at offsite bldg: http://www.garlic.com/~lynn/vmhyper.jpg tried to then get the software support released to customers ... but there was group in POK playing with some fiber stuff and they were worried that my software would interfer with getting their stuff released ... so they got it block ... their stuff eventually is released as ESCON. recent posts mentioning the effort: http://www.garlic.com/~lynn/2012c.html#23 M68k add to memory is not a mistake any more http://www.garlic.com/~lynn/2012d.html#19 Writing article on telework/telecommuting http://www.garlic.com/~lynn/2012e.html#52 M68k add to memory is not a mistake any more http://www.garlic.com/~lynn/2012e.html#54 Why are organizations sticking with mainframes? http://www.garlic.com/~lynn/2012i.html#74 HELP WITH PCOM - PASTE OPTION NOT WORKING CORRECTLY http://www.garlic.com/~lynn/2012k.html#3 printer history Languages influenced by PL/1 http://www.garlic.com/~lynn/2012k.html#47 T-carrier http://www.garlic.com/~lynn/2012k.html#48 1132 printer history http://www.garlic.com/~lynn/2012l.html#25 X86 server -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN