/>
X
Innovation

Staffing up for the cloud future

The last paragraph of my last blog post glossed over perhaps the most important part of the change in Microsoft's data centre strategy - the fact that it's hiring a new cadre of cloud data centre engineers with internet scale experience, bringing in staff who've been at companies that were built at cloud scale from the start; from Amazon, from Google, and from Facebook.

The last paragraph of my last blog post glossed over perhaps the most important part of the change in Microsoft's data centre strategy - the fact that it's hiring a new cadre of cloud data centre engineers with internet scale experience, bringing in staff who've been at companies that were built at cloud scale from the start; from Amazon, from Google, and from Facebook.

As data centres move from being arrays of servers running applications, to Infrastructure-as-a-Service utility computing platforms, the way things work changes dramatically. Each individual server stops mattering, and when they fail - as they will - load just migrates to another chunk of compute somewhere in the network. It doesn't matter where that load goes, it could be the machine next door, or a machine in another continent. All that matters is that the end user sees their service continuing…

It's an approach that demands a fundamental change in the way Microsoft delivers IT. The teams that used to maintain individual servers are now managing whole services, blending infrastructure, operations and applications. Task that used to take days to plan now need to be executed in seconds (and while automation helps, the administration team needs to remain in top of things). Clouds are a fundamental shift in data center operations - and with services needing to respond in Internet time a rollout can't be shifted a few days just because it's the Friday evening of a long weekend. Service users have become service consumers, and they're expecting giants like Microsoft to respond in the same ways as the Facebooks and Twitters of this world - giving them what they want when they need it, not in six months or even in two weeks.

Infrastructure as a service is a huge change from the old way of doing things. Instead of preconfiguring and testing servers on the bench, they're just dropped into racks as bare metal ready to PXEboot up a hypervisor and run whatever workload is needed. There's no building a custom OS for each box, no need to fine tune applications. With platforms like Azure the base hypervisor boots up in seconds, and then management tools deploy appropriate OS and service images to the newly live server, offering end users platforms.

So there's a lot of sense in Microsoft hiring staff with cloud data centre experience. As old services move off their dedicated hardware onto new Azure platforms, taking advantage of the cloud's distributed compute utility, Microsoft is going to need a cadre of engineers who can train and mentor existing staff in the new ways of working. Even so, the changes are going to be painful. There won't be the need for the same level of infrastructure engineering - and engineers will need to either learn new skills, or will end up having to leave or take lower-skilled posts handling basic data centre maintenance. Meanwhile, OS and application engineers will need to learn how to use the new management tools and to understand just how a distributed cloud platform works, producing and managing a catalogue of servers and services that can be used by anyone to deploy just the custom configuration they need - whether they're inside or outside the Microsoft firewalls.

Tomorrow's data centre doesn't need today's data centre staff - and certainly they don't need (or even use) today's ways of working. So it's good to see Microsoft reaching outside Redmond to hire in the skills and methodologies it needs to run its next generation data centres and services. After all, it's no good having a PUE of 1.05 if you're not actually taking advantage of it…

Simon Bisson

Editorial standards