This was the title of the presentation I gave at Grid World in Washington, USA, (sorry couldn’t resist adding the USA piece, after all it might have been in Washington in the north of the UK), and judging by the general nature of the event the question on everyone’s lips. It really reminds me of the early days of InterOp in the late 1980s when a hardened gang of technocrats knew that TCP/IP, as we used to call it, was what the world needed. The trouble was that the rest of world didn’t know it, and wasn’t too sure it was worth listening to the messages. Of course these days we can’t quite imagine a world with out IP and its ability to under pin almost every aspect of everything.
The game in those days was to find some one who had done something useful with TCP/IP and build huge cases on why this was a repeatable solution needed across the market. In fact it was just a niche, and it wasn’t until Unix really got going that the need for, and value of IP, really became clear. So it turned out in the end the business requirement wasn’t for TCP/IP at all, but that the something else, Unix, required TCP/IP to work. Sort of secondary level requirement, but of primary importance as something everyone could share, and build on, to deliver the functionality from their products that in turn met the business requirements.
And that in a paragraph seems to be the state of ‘Grids’, a random selection of case studies that are being presented as the justification of the need for grid technology, and a lot of people convinced that this is a breakthrough technology and frustrated others don’t see it. So where is the Unix to bring the undeniably right technology of Grid into general use?
I reckon there are two answers; the first is Smart Business Networks the developing business requirement to link members of a business ecosystem together to improve business reactivity and capability. The second is SOA which is being driven by the first as much as anything else. ‘No SOA with out SOI’ might be the war cry of the ‘Gridistas’. Actually it should be the war cry of the SOA side too. Exactly how do you cope with un predictable event driven activities which may originate externally without a new type of infrastructure, i.e. SOI? And what happens in a smart business network when an extended process crossing several companies needs to find the resources right now to support their services in the overall process chain?
That’s not virtualisation, an internal capability around the things you own control and manage, it’s a different game, and that’s where we need the Grid. Of course I could add a load more on fragmentation of requirements around the edge of a business making individual ‘allocations’ of resources and management a nightmare, but the basic principle of the big picture is clear enough. If you need proof look at the attendance list and see all the big technology vendors listed then look at the titles of the attendees, it makes the point!