Well no, to be fair no one from Monty Python actually said that, but the idea keeps hanging around, and it might actually have some merit after all. Microsoft is the latest big name to have gone for the idea of the datacentre in a container, with one on show at its recent Professional Developers’ Conference. While the idea is certainly not new, there is a new dimension to the idea that could have some far-reaching implications.
There are a number of big name vendors in the currently small market for these systems, including IBM, Rackable Systems, Digital Realty Trust and Sun Microsystems. With Microsoft in play as well, two and a half of the players have a solid track record in applications software and, most notably, database systems. The half is, of course, Sun, which is still the subject of a stalled acquisition by Oracle. Though US regulators have given the acquisition a clean bill of health the European Union equivalents have been far less sure.
But recent undertakings from Oracle to the EU about the independence of MySQL – the open source database it acquired a while ago – look likely to satisfy European misgivings, unstalling the acquisition. That would give Oracle its own hardware muscle as well as database systems well-placed to cover both the high-end and low/mid market sectors.
Couple this with the fact that these players, all except Microsoft for the time-being, are working together to develop standards for the datacentre-in-a-can business and things get a bit interesting. This is particularly so in the short term for any company that suffers the Cloud-phobic fear of not knowing where its data might be. Longer term, standardised `cans’ could have an indirect impact on much of the Cloud infrastructure.
The key to this is that word, `standardisation’. While it is often said that innovation is stifled by the application of standards, much of that innovative effort achieves precisely zero in the great scheme of IT things. The IT world is full of heroic technology failures – brilliant bits of technology innovation that failed because, no matter how clever, they missed the market in some way.
There is, in fact, far more innovation, particularly in the area of actual usability and fitness for purpose, when standards are in place. One just has to link the phrases `PC architecture’, `Intel x86 architecture’ and `Microsoft Windows’ together to see what I mean. Technically that basic triumvirate now stretches from the desktop to the largest racks in the biggest datacentres – they are, by and large, the building blocks, the very water droplets, of the Cloud. Yet those boring, hackneyed old standards have generated unceasing innovation in how the Cloud is exploited.
So, standardised ranges of standardised datacentre `cans’ running the pretty much standard options on applications and database systems could be very tempting for any company that either likes to have its data and systems where it can see them, or has a need for some real extra grunt for specific projects. Yes, that is what the Cloud can also offer but it does have weaknesses, particularly if latency in the processes being run is an issue. Quite often, having the processing grunt physically close can be a beneficial, indeed crucial decision. And because it will be standardised it will, for the time it is required, be a cost-effective option.
With the development of such standards there comes the wider implications for Cloud service providers. Some that provide datacentre services are already building new suites to standard specifications, so installing and running standardised IT infrastructure and software becomes an even easier task. To date it can still take a couple of months or more to get a bare datacentre suite populated and in production, but with as much as possible – particularly on the base software side – standardised (and probably pre-installed) this time can certainly be cut.
And while standardised software sounds anti-innovative, I suspect the opposite will be true. With the technology standardised - so its issues can be `forgotten’ – all that creative brain power can be applied to the development of new services to exploit that infrastructure…….in much the same way car design improved rapidly once the layout of steering, accelerating, braking and gear-changing controls had been settled upon.
Post new Comment