DYNAMIC INFRASTRUCTURE MODELS

CHAPTER FOUR: DATA CENTRE BURSTING

‘Inflatable competence.’

That’s what I've been calling it. Well, this week…

At a recent VMware User Group (VMUG) event, VMware Chief Cloud Technologist, Joe Baguley, shared how he explained data centre bursting to his mother:

‘It’s like a taxi that can get bigger when there are more people to get in.’

I thought that was brilliant - once I'd substituted 'taxi' with 'service capability' and 'people' with 'people with computers'!

Today's blog entry is not a sharing of experience but instead a warning. Being part of F5’s cloud practice means that, among other things, I must read a lot. Reading industry buzz ensures that:

a. I'm validating my ideas (it’s a rapidly changing industry) and

b. I see patterns. The largest pattern being what inspired this Dynamic Infrastructure Models series. This pattern being a fundamental flaw in almost ALL 'bursting' reference architectures I've read. The limitation of the network.

What's dynamic about waiting for a network change to reflect a service requirement?

Some solutions allow administrators to add more servers to a service, providing increased processing power on demand. But, unless the network can handle the decommissioning of a server from the service, when a reduced capability is required, you don't have network agility. You lack data centre symbiosis.

When John Burdette Gage, the 21st employee at Sun Microsystems, coined the phrase, "The network is the computer": this is what he meant.