I started at ICL on 9 January 1978 and have worked there ever since, although as you will see it's not been known as ICL since April 2002.
The Mainframe Years (1978-82)
A 2980 mainframe computer
My first job was Systems Software Support Engineer in the Manufacturing Division at West Gorton, where we ran ICL's flagship operating system VME (Virtual Machine Environment). In those days both mainframe hardware and software of the early "P" series mainframes range - 2970, 72, 76, 80, 82 - was in early development. Pre-dispatch trials had to be run (typically for 6-12 weeks depending on the configuration) to ensure the stability of the system. Very large configurations such as those for government and utilities often ran for months. It was a far cry from today's drop-shipment of systems direct from assembly line, with Mean-Time-Between-Failure averages measured in thousands of years.
A bank of EDS200 disk drives
It takes five of these to hold 1GB
The first mainframe machine I worked on, a 2970, was a real whopper: a whole 4MB of mainstore (requiring two cabinets - one for the store access controller and one for the memory chips - each cabinet the size of a single wardrobe and each chip holding only 1kb of memory) and four 200MB exchangeable disks each the size of a dishwasher. Once assembled, the machines ran engineering test software and later, when proven capable, a series of repetitive benchmark tests on VME/B development releases known as 5X23 and 5X27. Along with about 20 other engineers, I was responsible for general systems installation, configuration, support and the first-line diagnosis of system and VM crashes.
At first trials jobs were run from card packs. Later we executed them from data files. Even later, an exceptionally gifted colleague of mine conceived an elegant control mechanism that stored job set definitions as compiled modules, managed the execution of these jobs, compared their outputs to master copies held on disk, deleting all output files when checked, and maintained detailed log information in bespoke logs viewable from the master operator console. As one of only three people in the group to show an interest in software development, I designed and wrote the automated housekeeping, volume utilities and journal analysis "subsystems" of this new system.
In 1981 mainframe manufacturing moved to purpose built factories outside Manchester and shortly afterwards we began trialling a new range of machines based on more powerful processors and redeveloped peripheral architectures. When completed in 1982, the trials jobs control software allowed an entire pre-dispatch trial to be run on these "S" series machines using two commands - START and STOP. Having spent a few years diagnosing system crashes as far as the failing line of code, I was offered a job in the VME Product Support Unit and in July 1982 I moved to ICL offices in Staffordshire.
The Third-line Support Years (1982-89)
In my new life, I initially took responsibility for job control and scheduling areas of the operating system, later expanding into various other subsystems in both Upper and Lower Director. This was around the time of VME Release 6.02. I must confess to being a bit star-struck in those days. For a lad from the "sticks" (anywhere outside mainframe development) to be hob-nobbing with the very people who designed and wrote VME, people whose initials I'd only been fit to read off a piece of Project Log fiche a matter of weeks earlier, was a little overwhelming. But I soon got over it ;o)
The VME Product Support Unit office
Shortly after joining the Director Support group, I took over as team leader of the 4 man "Jobman" team, and between 1983 and 84 I moved on to lead the entire Director group of 19, including 5 home workers. Around this time, the third generation of modern ICL mainframes was nearing the end of development - an entirely new hardware architecture based on fibre-optic data transfer in-cabinet, with processor chips designed in Manchester and fabricated by Fujitsu. To meet the release deadlines we were working 12-hour shifts. Along with redeveloped hardware and a major rewrite of VME, there was a significant change in the worldwide support route. Once the new systems (now badged "Series 39" by the marketeers) hit the field, the support community began to receive system and VM dumps on tape. In a 28 hour stint over two days, I designed, wrote and tested a control procedure to simplify the access to these dump files for support technicians on-line that leveraged the new Series 39 dump formatter written by a colleague. I also wrote a series of script files to create a form of automatic dump processing. This was subsequently used for over ten years as the basis for the first level "pre-scan" of all Series 39 system crashes.
Series 39 represented a huge churn in the base: over 25% of the OS was new and inevitably there were problems in the early days. As the VME bug backlog continued to grow through 1985, I played a major part in the coordination and execution of various "bug eradication" exercises, on several occasions resolving more than 50 incident reports in a single week. In mid-1986 the VME Product Support Unit (along with the entire development team) moved back to Manchester from Staffordshire. In the Spring of 1987 I was one of three diagnosticians - one from each of the three operating system disciplines - selected for an ambitious bug clearance program of all fault reports that had been raised on earlier releases. Working alone on the "Director" faults, in the six weeks of the "Spring Offensive" I reduced the backlog from 256 to 8 and raised more than 50 software repairs.
In the period between July 1982 and the end of 1988 (when I stopped working full time on third-line support) I had maintained the highest average fault closure rate and raised more than 500 repairs. I now started to withdraw from mainstream support and concentrate on strategic issues with the support process. In 1989 I finally left 3rd line support to join the VME strategy group.
The Unix Years (1990-96)
After an abortive six months in the Strategy unit, where I struggled with trying to design changes to VME to make support easier when there was no allocated budget for any development to support those designs, in early 1990 I was approached to join the embryonic X/Open project. The ambitious aim of this project was effectively to implement Unix on VME - known internally during its development as VME-X and later when the marketeers got hold of it, as OpenVME. If successful ICL would be the first mainframe manufacturer in the world to be certified compliant with the X/Open standards for Unix system interfaces. One of the most pressing changes required to give VME a Unix feel was the introduction of personal authentication. In common with most mainframes of the time, "usernames" were shared between members of a workgroup. Usernames giving individual privileges were not available. This became my first "real" design - changes to the system logon software to support personal authentication.
Around this time the first PCs were being introduced into the office and I was one of the first in the X/Open project to get my hands on one. A whole 4MB RAM and 40MB of disk space! Did you notice that after 12 years in the industry, the PC on my desk now contained the same amount of memory as the first mainframe I worked on? During my years with X/Open project, I worked on the design, implementation and testing of the X/Open Transport Interface (XTI) for VME-X and later the General Terminal Interface (GTI, that is, Asynchronous Terminals). I also designed and implemented a facility known as connection offering, which allowed listening processes to handle a virtually unlimited number of connections. ICL did indeed achieve the first certificate of compliance for a mainframe system from the Open Group.
My final project on OpenVME was the porting of a major office automation application (email, calendaring, bulletin board) originally developed for Unix. This took a couple of years and was successfully deployed in-house. Towards the end of the development cycle I was offered the chance to retrain on Microsoft Technologies.
The Microsoft Years (1996-2001)
For eight weeks in October and November 1996, I immersed myself in Microsoft training, passing all four exams for my MCSD accreditation at the first attempt. For the record, these were Windows Operating System Architecture (WOSA) I and II, Access 95, and Visual Basic 4.0.
Microsoft accreditations "expire" after a period of time (usually 2-3 years), so I am no longer officially an MCSD. In the first few years of ICLs "Microsoft Partnership" retraining and reaccreditation was encouraged but somehow I never found the time. Now the focus is more on internal certification and for reasons that will become clear later, Software Development accreditation is not uppermost in our minds.
John & I at the Parthenon the morning before
flying back to UK
While some of my colleagues (mainly those with a Unix background) continued to have a preference for developing in C, I trained on and exclusively used Visual Basic. At the time within the industry there was an ongoing (not to say interminable) debate on the relative merits of VB, Visual C++, Delphi and Java. As usual with this kind of debate there was a lot of evangelising and squabbling. As time went on there became less and less that you couldn't achieve with VB. Certainly after working extensively with VB versions 4, 5 and 6 over a period of four years I never found anything I couldn't do with it simply, quickly and effectively. But as my group eventually agreed: a launguage is a language is a language. Use the one you're most familiar and comfortable with. In most cases it doesn't matter. (Portability requirements excepted)
One of the first projects I worked on after qualifying took me to Greece (Athens, to be precise) seven times in the course of a year. Mostly, it was very hard work: long hours, sometimes in server installations miles from a decent restaurant ;o) On one occasion though, my colleague and I finished unexpectedly early one evening due to a fault in a different system, and were due to fly back the next day. We searched out a new place to have a meal and a drink and happened upon a tavern in the back streets of Athens, unaware that a floor show would be starting around 10pm.
The bazouki players we could have done without, but the belly dancer was a treat. Not all business trips are created equal! And if you think I look embarrassed here, you should have seen my mate's face when the dancer grabbed hold of his head and held his face against her wildly gyrating belly!
In 1998 I started concentrating my work around SQL Server. By dint firstly of attending the Tech.Ed 97 conference in Nice and following the SQL Server thread, and secondly of returning from Nice to give a presentation that impressed the Chief Engineer, I found myself being treated as ICL's SQL Server "expert." Never has my favourite definition of expert been more apt: 'ex' as in 'has been' and 'spert' (sic) as in 'a drip under pressure.' Still I made the best of it. Amongst other things it led to me making the pilgrimage to Seattle twice, so it wasn't a complete loss.
But the writing was on the wall for application development in our division. The market was changing. For about a year, I busied myself learning about Enterprise Application Integration and the early versions of MS BizTalk Server. Once again this resulted in some interesting trips: to Nice again for the BizTalk Airlift in early 2001; and shortly after that to Belgium for a 3-day training event. EAI was a hard sell though. Hard to explain to customers exactly what it was they were getting for their money. In the end, very little business was generated and at the end of 2001 I transferred to the unit responsible for providing computing services to the Home Office, thus narrowly missing a round of redundancies in the unit I left behind.
The Infrastructure Years (2001-present)
While I trained as an MCSD, many of my colleagues had followed the MCSE (Microsoft Certified Systems Engineer) path, and it was this expertise which became flavour of the month as the business moved away from bespoke application and product development and towards the provision of systems and infrastructure services. This was a steep learning curve for me. One which I'm still on, if I'm honest.
In April 2002 the rebranding of ICL as 'Fujitsu Services' was completed. ICL signs were taken down from all the buildings and the staff divided between the services and consultancy divisions. After over 24 years of working with the same company, it felt very strange still to be working for that company, but be called something different. Even today, it's hard to refer to my career prior to 2002 without stumbling over some clumsy concatenation of business identities: "How long have you worked here?" ... "Well, with ICL-stroke-Fujitsu since 1978."
At about the same time as the rebranding, I was asked to head up the technical team responsible for refreshing the infrastructure and applications at the Home Office. This must rank amongst the most challenging tasks I've ever undertaken, and was to take the best part of three years owing to the extreme complexity of the application set and the size of the estate.
Summing Up
I've never been a planner. I've never said to myself "by 'this' time in my career, I want to be 'here'." But if you take that unplanned approach to life, you have to be prepared to go with it; roll with the punches; take the rough with the smooth. There have been times in my career when I've been on top of the world: working with great people, for great people, and achieving great things. Then there have been times when I've felt as if I'm marking time, not really achieving much, and in the company of...well, let's just say people who are not so great. I don't think that's different from any other walk of life and I'm not sure I would change any of it if I could. All in all it's been, and continues to be for the most part, a blast.
But I do experience feelings of frustration when I hear people say: "28 years at the same place? You must be institutionalised." Yes, that was actually said to me by some spotty erk at a recruitment agency during the most recent round of redundancies when we were all "considering our positions." Well let's see. 28 years is a long time to stay in one place, doing one job. But have I really been doing that? Since 1978 I've worked in five different permanent locations (two of them more than once), visited more than ten other locations on a regular ad-hoc basis. I've worked on over twenty different types of hardware including mainframes, midrange and PC systems, running VME, Unix or Windows, and I've successfully held down jobs as a support engineer, operating system developer, tester, integrator, I've ported applications and developed them from scratch, written software patches, liaised with customers, managed teams of people from 1 to 20, planned projects, run special interest groups, presented at Microsoft Developer Days, attended the Executive Briefing Center, run workshops, presented to the ICL Technical Council, designed an infrastructure to support 20,000 users and I'm a Fellow of the British Computing Society. So no, I don't feel "institutionalised," thanks.