My ramblings on the stuff that holds it all together
I’m mainly interested in Martin’s last comments,
“Virtualization can be a real business enabler, a way of providing the IT infrastructure in a more energy efficient way, as well as allowing the business to have a more on-demand, fluid type infrastructure. One of the ideals would be when we can get to a follow the sun model, an infrastructure that runs with your business around the globe following where the energy is cheapest at that specific time, with a fluid type failover. That I could have my New York teams run from London whilst the server guys update the New York infrastructure is the kind of evolution of the platform that could be a real enabler, the investment, the planning and the applications all need to be in line with the infrastructure, as does your business processes, the charge back.”
That would be cool, and it’s mostly achievable in the world of replicating storage and abundant high-bandwidth connectivity between major locations, failover to the other side of the world and it gives you a good sized window to upgrade/maintain the underlying platform, even replace it.
I’ve already achieved a similar thing with VI3 customer installations for a test/dev environment. no real reason why this couldn’t be the same for a production platform and on a global scale.
Ops want to be near the kit they support for that touchy-feely warm feeling? sure, provision the VM layer and you can have it in your DC for the support hours you cover… we’ll even shift it somewhere else when you are celebrating Christmas, Thanksgiving, Talk like a pirate day etc.
Same principal for web-type eCommerce services, virtually move the supporting infrastructure to be physically nearer the user base (lower network latency, distance etc.) during their peak hours..
Removing the Software<->OS<->Hardware dependency can only be a good thing, and with all this web 2.0 stuff people want stuff quick, that doesn’t work in the physical world where I have to order, deliver, rack, cable, install something physical to provide a new service – people want it now and the only way to do this is using virtualisation – better to provide a “grid” and allocate resources out of it as required – you can easily add capacity as you go in a controlled/planned manner.