“The Computer’s crashed” – No it hasn’t!

Much misused terminology now in general use and referring to an unexpected hard stop in the operation of desktop or hand-held computing device hardware or software, a mobile phone or even the technology fitted to your car.

Background: OLS Computer Services (UK) Ltd. (OnLine Systems Inc.)

During the 1970s and 80s, Online Systems Inc. was a large time-sharing service bureau business headquartered in Pittsburgh PA. In 1978 the company acquired Leasco Response Limited, a computer firm situated in a prestigious office block in London’s Knightsbridge. This created OLS Computer Services (UK) Ltd., said to be Europe’s first time-share service bureau (today’s cloud and data centre).

In autumn 1976 OLS moved to Southgate Road, Islington just north of the City of London. The firm took its four ageing Hewlett-Packard HP-2000s and also installed a Digital Equipment Corporation DEC PDP-11 front-end processor. The latter acted as a gateway for UK and European-based customers to access Online System’s business applications located on one of sixteen Pittsburgh-based DECsystem-10s.

OLS sold packaged business services - today’s desktop applications and mobile device-based ‘apps’ - including payroll, accounts, inventory management and the like, running on the aforementioned HP-2000 TSB time-sharing computers. For research and academic work “raw time” was sold to universities at £750/hour (over £10,000 per hour accounting for inflation).

The customer (“end-user”) accessed services via either a GPO (The Post Office) type 2 dial-up modem (see separate post) or a rotary-dial telephone connected to an “acoustic coupler” (the latter operating at the grand speed of 110, or later 300 CPS). This 1960s communications technology has thankfully been replaced by broadband, Wi-Fi and GPRS. User access was enabled via a hefty Data Dynamics 390 or a Westrex 33 punched paper-tape enabled teletype terminal (I can still hear the clatter) and later Haseltine 2000 “glass teletype” terminals (aka: Visual Display Units – or VDUs).

In 1978, the British engineering and consulting firm WS Atkins - now the UK’s third largest engineering consultancy business - made a move into the bureau services arena by purchasing OLS (UK) Ltd. renaming it Atkins-Online Ltd.

But this really was a “crash”

OLS ran a five-day, three-shift rota with weekend emergency standby cover for both operations and engineering. The only period of unattended operation was between the hours of midnight and 6.00am. Overnight sometime in late 1978 or early 1979 one of the HP-2000s’ Control Data Corporation CDC 854 (aka: ‘EDS-8’) disc read/write head arrays dropped onto the rotating disc surface. The failsafe design meant that the entire head unit quickly retracted to its home position away from the fast rotating disc to avoid damage.

But, as the read-write head dropped, it twisted and became stuck touching the disc surface – this then forced all other read/write heads on the same array to dislodge remaining in position moving across the entire surface of all inner platters. These then continued to rotate at 2400rpm continuously for another 7 hours.

Leaving a router bit set against a fast rotating aluminum billet and then going home for the day springs to mind!

At 6.00am the day-shift operator ‘opening up shop’ was presented with a highly commercially damaging situation – the entire computer room was filled with fine metallic particles. The disk drives were fitted with multiple filters and fans to trap and move the odd small particle but were not designed for the overwhelming quantity of airborne aluminium debris. Amazingly, all four HP-2000s remained in operation, barring the one now disabled disk drive which proved surprisingly robust in that it never stopped rotating all night. All systems required powering down for investigation so the entire operation was halted. 

Recovery: more disgruntled customers

With all four HP2000 systems down (five including the PDP-10) the main priority was to restart normal services as soon as possible but first the engineers needed to assess the damage. Not surprisingly it transpired that - apart from the obvious need for a replacement disk platter and head unit! - every boxer fan and filter in the entire computer room needed either cleaning or replacing. The engineers created an alcohol bath - soaking the parts for over two weeks - in an attempt to resurrect them for re-use but many were eventually replaced. The air-conditioning and fire protection systems also required careful inspection and maintenance.

By 4.00pm on ‘crash day’ the operations team, which I headed, managed to get one HP-2000 operational so we moved all live customers across. (Customer Experience wasn’t an art form back then but suffice it to say that our customers weren’t best pleased - nothing’s changed here then!).

It took us most of that month to get the other three HP-2000s in good shape with customer accounts and applications re-loaded. The PDP-11 was no trouble, it rebooted straight away (well, it was basically only a rack of circuit boards connected to a DecWriter terminal).

Conclusion: we’ll never know!

We never did discover a cause for the catastrophic failure. It was said to be most likely due to a fault developing on a single read/write head logic card – this caused the array to became loose and after contacting the spinning disk it twisted becoming stuck between the platters.

So, what happened to that disk pack?

The infamous disk pack became a trophy - it was dismantled and individual platters distributed across the engineering and operator teams. Three, showing a level of damage that clearly caused the disc to reach temperatures causing warping, remain in my possession (see photos).

Even today when I hear or read talk about computers “crashing” it brings to mind the fact that computers don’t crash (nor does software – it halts) and then I look fondly at one of my trophy discs.

And, if you’re interested in how mini and mainframe computer based time-sharing services: paved the way to the future...

By allowing a large number of users to concurrently interact with a single computer, “innovative” time-sharing dramatically lowered the cost of computer access, making it possible for individuals and businesses to access a system without the CapEx, and promoted the interactive use of computers and the development of interactive applications. Although the DEC PDP-10 is said to be the machine that made time-sharing common in the 1980s this somewhat forgotten major technological shift in computing history began in the 1960s’ mainframe era and came into commercial prominence in the 1970s with the HP-2000 series (and others) starting what has morphed into today’s ‘cloud’ access via ‘IP’ (TCP/IP, packet switching – how the internet transmits stuff!).

#Hard Disk Crash #Hewlett Packard # DEC PDP-10 #Leasco Response # Online Systems Inc #OLS Computer Services #Rob Sherwin

© Rob Sherwin 2019: Chief Operator, OLS Computer Services (UK) Limited.