Original URL: http://www.theregister.co.uk/2011/05/24/tom_west_obit/
Data General's Tom West dies
The man who put the soul into a new machine
Obit Tom West, who created Data General's Eclipse 32-bit mini and was immortalised in Tracy Kidder's Pulitzer Prize-winning book Soul Of A New Machine, has died.
Credited with helping to save Data General (DG) after DEC announced its VAX supermini in 1976, Joseph Thomas West III was born on 22 November 1939 and died at his home on 19 May 2011, aged 71. It is thought that he suffered a heart attack.
West was the son of a business executive, and the family moved quite often - he attended four schools and then studied at Amherst College.
He was a folk singer towards the end of the 1950s and worked at the Smithsonian Observatory in Cambridge, Mass, before returning to Amherst and gaining a bachelor's degree in Physics. He continued working at the Smithsonian, going to other observatories and ensuring that the time was precisely synchronised.
West then joined the RCA corporation and learned about computers, being largely self-taught, and then joined Data General and worked his way up the engineering ladder.
DEC shipped its VAX 32-bit supermini in 1978. This was in the era well before Intel's X86 desktops and servers swept the board, when real computer companies designed their own processors. The 16-bit minicomputer era had boomed and DEC was the number one company. DG was the competitive number two sometimes known as 'the bastards' after a planned newspaper ad that never ran, and was a Fortune 500 company worth $500m. But 16-bit minis were running out of address space (memory capacity) for the apps they wanted to run.
The Eagle skunkworks
DG launched its own 32-bit supermini project known as Fountainhead. It wasn't ready when DEC shipped the VAX 11/780 in February 1978 and suffered from project management problems, so it is said. West, far from convinced that Fountainhead would deliver the goods, started up a secret back-room or skunkworks project called Eagle to build the Eclipse MV/8000, a 32-bit extension of the 16-bit Nova Eclipse mini.
He staffed it with an esoteric mixture of people, some of them recent college graduates, and motivated them not with cash, shares or external incentives but by the sheer difficulty of what they were trying to do. It was described as pinball game management. If you got to succeed with this project or pinball game the reward was that you got to work on the next, more difficult pinball game.
Towards the end of 1979 it became clear that West's team was going to bring its project to fruition before Fountainhead. DG customers were increasingly buying DEC VAXs and DG, running out of cash, experienced tense internal competition as the Fountainhead and Eagle development teams fought for limited development funding. The 16-bit Eclipse was a problem-strewn product and its reputation was bad. Nevertheless West's decision to extend the Eclipse architecture and his intense management style proved to have the edge over Fountainhead.
West's skunkwork staff knew that were working at the frontiers of supermini development, and the organisation, atmosphere and personalities of the protagonists were captured by Kidder in his book. One of the most memorable characters was the man responsible for the microcode who produced nothing for weeks and then, in an outstanding burst of concentrated creativity, produced it all in a very short time.
The MV/8000 was launched in 1980 and fuelled a turnaround in DG's revenues. It was hailed as miraculous and West became an engineering god who could do no wrong. Unfortunately, he never again achieved this level of success, and the Eagle project proved to be his finest work. MV/8000 sales roared ahead and DG passed a billion dollars in annual sales in 1984.
But the workstation era, started and boosted by Sun and Apollo, passed DG by. So too did the PC, with the DG-1 PC being a poor product, and the firm on a downwards track. Microcomputers entered the supermini market - remember MicroVAX? - and DG found it could no longer afford to develop its next-generation processors, not having the revenue strength of DEC.
Killing his own creation
West helped author a report in 1988 recommending that DG should develop hardware systems using commodity microcomputers, instead of building its own proprietary processor, to run Unix better than anybody else. In other words he was responsible for giving life to the MV/8000 and now proposed killing it. DG CEO Edson De Castro agreed with the proposal and the MV line was run down.
A Motorola 88000 product line called AViiON was started up; it ran Unix and therefore all Unix apps that could run on the 88000 processor. Some say that West saved DG's bacon a second time by killing off his own MV/8000 creation and moving to the 88000.
Over time it became apparent that the 88000 was not succeeding against Intel's X86 line of processors, and Motorola killed it off. DG transitioned to the X86 but wasn't able to differentiate its hardware enough and sales revenues fell.
A high-availability storage array was created to run with it the AViiON, and then with other DG computers. It became branded CLARiiON, and was the reason EMC bought the much weakened DG in 1999. CLARiiON arrays have survived to this era, only being replaced by the VNX  line earlier this year.
Missing the networked desktop
From a personality point of view West was seen as aloof, passing people in DG's corridors without acknowledging them. It's been said that this attitude permeated DG and it wasn't that good at internal communications. West was seen as a rackmount minicomputer and then microcomputer man. He would be called a server guy today. Back then it seemed he thought workstations and networks, other than networks bringing clients to his MV/8000 and then AViiON systems, were minor irrelevancies, and destined to stay that way.
So DG didn't seriously get into workstations, even though one El Reg correspondent said: "DG developed an 88000-based DG-UX workstation codenamed Maverick that blew the doors off the then-current flavour of Sun, but Tom had it killed."
He was also thought to have a less than stellar record with networking - this was long before Cisco was rampantly dominant - and DG could have had a router offering to go with its bridges, terminal servers and hubs, but it couldn't settle on a protocol stack, TCP/IP then not being dominant. No DG router emerged, and one view is that West effectively had DG's networking products killed off and thrown away.
DG development teams not favoured by West became, it's said, quite bitter about being kept outside the magic circle. This has been said about the workstation, networking, disk, tape and printer groups. But times were hard at DG, with development funds in short supply, and West bestrode the development effort like DG's in-house colossus. His computer projects got the lion's share of the funds. Regrettably, though, he did not see the era of client:server computing and distributed desktops and that blind spot helped cause DG's demise.
Our correspondent said: "There were a lot of 'Tom Wests' in the Massachusetts minicomputer world toward the end, and not all of them worked at DG - or DEC. The failure to see that that which had worked marvellously well for 20 years was now irrelevant was endemic to the industry, and the people who couldn't see beyond their noses killed the whole Rt. 128 scene dead."
West saved DG
The straight truth is that "without Tom West driving the development of the MV/8000 there would have been no MV/8000, and DG would have died altogether circa 1980." Because of Kidder's book West became, for a while, the most famous computer engineer in the world.
He deserved that accolade and, with the MV/8000 skunkworks led a group of people, many of whom would have been classed as no-hopers, to produce a machine absolutely crucial to DG's future. It was his finest hour and a triumph. Read Kidder's book if you can find it, if only for nostalgia. We won't experience those days again.
West retired in the late 1990s and spent part of his time sailing. He is survived by two daughters, two ex-wives and a sister. ®