[email protected] (Anne & Lynn Wheeler) writes: > VNET wiki reference > http://en.wikipedia.org/wiki/IBM_VNET > > misc. past posts mentioning internal network > http://www.garlic.com/~lynn/subnetwork.html#internalnet
re: http://www.garlic.com/~lynn/2013b.html#7 Spacewar! on S/360 warning: vnet topic drift. JES2 had inherited some networking support from HASP that had "TUCC" identifier in the source statements. It used spare entries in the (255 entry) psuedo spool device table ... for network defintions ... typically able to define 160 or so entries. However, 23Jun69 unbundling announcement started to charge for application software (company did make the case that kernel software would still be free). JES2 had fairly heavyweight development and pricing policies required that monthly price cover the ongoing costs plus the upfront development costs. Even with inherited lot of networking support from university hasp ... there was no forecasted price for JES2 networking that covered its cost (company normally did three forecast levels, high, medium, and low ... assumption was that number of customers increased as price went down ... but there was no price times number of customers ... that covered the upfront JES networking costs). VNET had modern layered architecture (compared to JES and other of the period) ... had no limit on nodes and also could support "drivers" that talked to other infrastructures. At the time, the internal network had more nodes than could be defined in JES ... so the basis for the internal network was all VNET ... but VNET did have drivers that could talk to JES as boundary nodes (JES couldn't be trusted at other than boundary ... since it would trash traffic if either the origin or destination node wasn't in its table ... even at boundary, JES would trash traffic where that JES was the destination ... if the originating node wasn't in its table). JES lack of clean layered architecture also resulted in traffic between two JES systems at different versions would crash the MVS system. This became increasingly common as the internal network started to pass 1000 nodes world-wide. As a result, protocol conversion routines were added to the VNET JES drivers ... which would convert from any JES format to the specific format required by the specific JES version that it was directly talking to. This http://en.wikipedia.org/wiki/IBM_VNET makes mentioned of 19.2kbit trans-atlantic satellite circuit. This contributed to the infamous case of traffic from San Jose JES system crashing Hursley MVS systems (because the releases were at different level). The crashes was blamed on the Hursley VNET system ... because they had failed to start the correct JES driver that converted to the format required by the Hursley JES release. In any case, this was after the FS failure and the mad rush to get products into the 370 pipeline ... as well as POK convincing corporate to kill the vm370 product, shutdown the burlington mall vm370 development group and transfer all the people to POK for MVS/XA (or otherwise MVS/XA wouldn't meet ship schedule nearly 7-8yrs later; Endicott managed to save the vm370 product mission, but had to reconstitute a development group from scratch ... online share archives has things to say about vm370 product code quality during the period). As a result, corporate wasn't approving the announcement and release of VNET as customer product. The JES group finally convinced corporate to allow VNET release/announce as part of a pricing gimick ... it would be announced as a combined JES+VNET product ... where JES & VNET were both priced the same (little obfuscation since VNET did have JES drivers). Then the combined JES+VNET forecast times the price was finally larger than the JES development costs (VNET costs being nearly negligible) ... as an aside, this wasn't the only time that the slight-of-hand happened using vm370 products to cover MVS product costs. misc. past posts mentioning hasp, jes, nji, etc http://www.garlic.com/~lynn/submain.html#hasp besides the story here about Edson talking to ARPA http://en.wikipedia.org/wiki/Edson_Hendricks the internal network was larger than the arpanet/internet from just about the beginning until sometime late 85 or early 86. I've pontificated it was partially because vnet had layered implementation with effectively ability for gateway in every node ... something that the internet didn't get until the great cutover to tcp/ip on 1Jan1983 (at the time arpanet was approx. 100 nodes and 250 hosts, while the internal network was rapidly approching 1000 ... which it passed that summer). again, past posts mentioning internal network http://www.garlic.com/~lynn/subnetwork.html#internalnet post with copy of vnet 1000 node announcement http://www.garlic.com/~lynn/2006k.html#3 and http://www.garlic.com/~lynn/2006k.html#8 above also includes other weekly new node announcements during 1983 as well as summary of all internal locations that had new nodes added during 1983. There were also posters and desk ornaments as part of the 1000 node event ... picture here http://www.garlic.com/~lynn/lhwemail.html#oldpicts one of the major reasons for number of internet nodes passing internal network ... was PCs and workstations appearing as internet nodes ... while the communication group was in its battle to restrict workstations and PCs to terminal emulation (doing its best to fight off distributed and client/server computing). misc. past posts http://www.garlic.com/~lynn/subnetwork.html#emulation -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
