Although the arguments over which was the first “real” computer rage on: is it the Zuse Z3, from Germany which was Turing complete but lacked conditional branching; is it the Colossus Mark 1, from the UK, which was programmable but not Turing complete; is it the ENIAC, from the US, which was both programmable and Turing complete but did not store any programmes onboard, or perhaps it was the Manchester Small-Scale Experimental Machine (or Manchester Baby as it was known), also from the UK, which was programmable, Turing complete and did store programmes onboard. Whichever is your personal choice, one thing is clear, they were eye-wateringly expensive.
How expensive? Well, in 1956, the Farranti Pegasus sold for 50,000 GBP, that was when the average house in the UK would set you back a mere 2,500 GBP. To put that in context for you, that would mean that today the average computer would cost you just over 3,000,000 GBP! Of course, back then, there was no such thing as an “average” computer. Not every household had one, in fact not even every business had one. To demonstrate just how rare they were, Ferranti only sold 26 Pegasus 1 and 12 Pegasus 2 computers – of course, that still made it Ferranti’s most popular computer.
So, if there were so few computers back then, how did computerisation work? Well, simply, you outsourced your computing entirely to something called a computer bureau. The bureau was there for companies that had neither the financial wherewithal to purchase their own computers, nor the expertise to program, operate and maintain such purchases. They worked by offering economies of scale; for a price, you could send your raw payroll data, for example, to a bureau and they would run the programs through the computer and return you the payroll output data.
Of course the downside of this was fairly obvious: you had to transport your data to the bureau; their computers had to, reliably, run the calculations – this at a time when an engineer had to carry out an hour of preventative maintenance on the computer every morning before anyone went near it! Finally, the finished output data had to be returned to you. All this had to happen without a single error, if any part failed then the whole failed and you were left without your data. Bad news if it was payroll data and you couldn’t pay your workers. Of course, actions in mitigation had to be devised – what would you do if you didn’t get your payroll data back in time? Well you could just pay last month’s payroll, as a temporary measure, and make the required adjustments in the following month.
Fast forward a couple of decades and the micro-processor comes on the scene. Computers plummet in price and every company who wants one can have one. They’re used for everything in the enterprise, from calculating the all important payroll, to the even more important managing the company’s lottery syndicate. Then something odd happens. Just as hardware is becoming simpler and cheaper we go and boik it all up by making the software more complex and more expensive. We add layer upon layer of abstraction; we have web servers talking to messaging servers talking to application servers talking to database servers and so on and so forth until the average company, not specialising in computing, throws up their hands and cries “Stop! Can’t I just outsource this into the cloud or something?”
And with that we’ve come full circle. Now, for a price, you can send your raw payroll data, for example, to a SaaS application in the cloud and it will do the calculation and return you the payroll output data. Of course, you may wish to have your own programs for your enterprise, but you might not want to be saddled with the cost of maintenance of the application and storage servers. Well cloud computing can help you there too. Just purchase compute time and storage facilities as you require, and spin up more servers over peak times, and spin them down again when you’re done. Maximum flexibility minimum cost.
Of course the downside of this was is fairly obvious: you had to transport your data to the bureau have to send your data to the cloud; their computers had have to, reliably, run the calculations – this at a time when an engineer had to carry out an hour of preventative maintenance on the computer every morning before anyone went near it! ISP can lose connectivity for days at a time. Finally, the finished output data had has to be returned to you. All this had has to happen without a single error, if any part failed fails then the whole failed fails and you were are left without your data. Bad news if it was is payroll data and you couldn’t can’t pay your workers. Of course, actions in mitigation had have to be devised – what would will you do if you didn’t can’t get your payroll data back in time? Well you could can just pay last month’s payroll, as a temporary measure, and make the required adjustments in the following month.
Wait a minute… isn’t this all starting to sound rather familiar? If, like me, you think it is – then maybe it’s time to put a Beatles track on the turntable and shout viva la ‘60s!!