jperr...@pacbell.net (Jon Perryman) writes: > * UNIX: TCP/IP was not publicly available until the 70's. Prior to > that, simple communications were available. > > * z/OS: SNA existed long before TCP/IP was available. SNA was a > robust, reliable and secure communications methodology. Once TCP was > became available, we had the same situation as Betamax versus VHS. TCP > won.
arpanet was host-to-host with IMPs from late 60s ... and in many ways similar to SNA (but well before SNA). big problem was that it wouldn't support large distributed ... and frequently autonomous, decentralized infrastructure ... and so start was made on internetworking protocol. the great change over of arpanet to internetworking (tcp/ip) protocol came 1Jan1983. at the time there was approx. 100 IMP network nodes with around 255 connected hosts. by comparison in 1983, the internal network was rapidly approaching 1000 nodes which it passed Jun1983 ... some internal network references for 1983 in this past post (in some sense it had gateway in every node which greatly simplified semi-autonomous expanding the network and was major factor in it being larger than arpanet/internet from just about the beginning until possibly late '85 or early '86) http://www.garlic.com/~lynn/2006k.html#8 other past posts mentioning internal network http://www.garlic.com/~lynn/subnetwork.html#internalnet note ... virtual machines, gml (morphs into sgml, html, etc), lots of interactive stuff ... all came out of the IBM cambridge science center ... some past posts http://www.garlic.com/~lynn/subtopic.html#545tech the internal network also came out of the science center, co-worker responsible http://en.wikipedia.org/wiki/Edson_Hendricks the internal network was not SNA (& not VTAM) ... technology similar to the internal network was also used for the univ. bitnet (where this ibm-main mailing list originated) ... some past posts http://www.garlic.com/~lynn/subnetwork.html#bitnet wiki reference: http://en.wikipedia.org/wiki/BITNET starting in the early 80s, I had a HSDT project with T1 and faster speed links ... supporting both internal network protocol and tcp/ip ... some past posts http://www.garlic.com/~lynn/subnetwork.html#hsdt one of the issues was SNA/VTAM only supported up to 56kbit links ... in the mid-80s, we were having some equipment built on the other side of the pacific. Friday before a trip, the communication group announced a new communication discussion group with the following definitions low-speed: <9.6kbits medium-speed: 19.2kbits high-speed : 56kbits very high-speed: 1.5mbits monday morning on the other side of pacific low-speed: <20mbits medium-speed: 100mbits high-speed: 200-300mbits very high-speed: >600mbits ... As part of trying to justifying only having support up to 56kbit links, the communication group prepared a report for the executive committee why customers wouldn't want T1 support until sometime in the 90s. As part of the report, they did a study of 37x5 "fat-pipe" support at customers ... multiple parallel 56kbit links treated as single logical link. They showed that the number dropped to zero around five or six parallel 56kbit links. What they possibly didn't realize was that telco tariffs for 5 or 6 56kbit links were about the same as single T1 link ... and customers would switch to full T1 and non-IBM boxes. At the time, we did a trivial customer survey of installed T1 links and found over 200. I was also working with various institutions and NSF ... and we were suppose to get $20M to tie together the NSF supercomputer centers. Then congress cut the budget and a few other things happened, and finally NSF released an RFP. Internal politics prevented us from bidding on the RFP ... the director of NSF tried to help, writting the company a letter (copying the CEO) but that just made the internal politics worse (as references to what we already had running was at least 5yrs ahead of all RFP responses). Some old NSFNET related email http://www.garlic.com/~lynn/lhwemail.html#nsfnet NSFNET backbone eventually morphs into the modern internet, reference http://www.technologyreview.com/featuredstory/401444/grid-computing/ along the way the communication group was spreading all sorts of FUD and misinformation (regarding NSF supercomputer backbone) ... some of the misinformation email was collected by somebody in the communication group and forwarded to us ... reference here (heavily redacted to protect the guilty) http://www.garlic.com/~lynn/2006w.html#email870109 In later part of the 80s, the communication group attempted a patchwork solution with the 3737 ... a box that supported T1 link ... but only had aggregate throughput of 2mbit/sec (T1 is full-duplex 1.5mbit/sec or 3mbit/sec aggregate, EU T2 is full-duplex 2mbit/sec or 4mbit/sec aggregate). Because VTAM line processing wouldn't keep the faster links busy ... the 3737 spoofed a CTCA to the host vtam and immediately ACKed the local VTAM transmission. The 3737 then had huge amount of buffering and non-VTAM line paradigm with remote 3737 trying to keep line running at full-speed. past posts with more 3737 details http:/www.garlic.com/~lynn/2011g.html#75 http:/www.garlic.com/~lynn/2011g.html#77 about the same time that the communication group was spreading FUD and misinformation regarding the NSFNET backbone ... it was also spreading misinformation justifying the conversion of the internal network to SNA/VTAM ... which required an enormous increase in allocated resources. http://www.garlic.com/~lynn/2006x.html#email870302 http://www.garlic.com/~lynn/2011.html#email870306 if there was to be any conversion of the internal network, it would have been significantly more cost effective and better performance if the internal network had been converted to tcp/ip ... similar to what bitnet did. late 80s, a senior disk engineer got a talk scheduled at an annual, internal, world-wide communication group conference supposedly on the subject of 3174 performance ... but opened the talk with the statement that the communication group was going to be responsible for the demise of the disk division. The communication had corporate strategic ownership of everything that crossed the datacenter wall. They were strenuously fighting off distributed computing and client/server trying to preserve their dumb terminal paradigm and install base. The disk division was seeing drop in disk sales as data was fleeing the datacenter for more distributed computing friendly platforms. The disk division had come up with a number of solutions to correct the problem, but was constantly vetoed by the disk division. This was significant factor contributing to company going into the red a few years later. other recent reference http://www.garlic.com/~lynn/2013m.html#100 SHARE Blog: News Flash: The Mainframe (Still) Isn't Dead -- 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