Gallery

Hyper-V is not hype

Posted on 3 Mar 2009 at 17:18

Jon Honeyball finds that hyper-v really is all it's cracked up to be, and David Moss gets to grips with installing an open-source cms on windows server 2008.

I've been talking to a number of chums who are architecting large-scale virtualisation solutions for their machine rooms and datacenters, and who have decided to go with Microsoft's Hyper-V. The results they're reporting are very encouraging indeed: to a man and woman they have nothing but praise for the capabilities in real-world stress-testing of Hyper-V. Of course, there have been some inevitable glitches and a few issues to sort out, but nothing that has caused them to wince and really wish that they'd gone with a competitor's product (whether that be VMware or Zen, for example).

Although I'm usually highly suspicious of manufacturer-supplied white papers, because you're never really sure whether you're getting the truth, the whole truth or nothing but the marketing spin, but there's one on the Microsoft UK website that caught my eye the other day and which will reward the reading.

Take a look at http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=4000003241 to read a case study that involves the move into virtualisation using Hyper-V made by Perth and Kinross Council, because it "wanted a more cost-effective delivery environment for its growing portfolio of citizen-centric services, some of which required multiple servers consuming large amounts of electricity". (We'll draw a polite veil over the advisability of the use of PC phrases such as "citizen-centric", if you don't mind.)

What I found so interesting about this study is that this council has managed to combine its move into the virtual world, with a simultaneous rapid deployment of Windows Server 2008 into a clustered solution spanning two main sites. The resultant money and power savings are impressive and speak for themselves: "The council forecasts that in the first year it will save £100,000 ($152,000) compared to the cost of buying new physical servers. Its carbon footprint has improved with annual power savings of 350,000 kilowatt hours of electricity equating to an annual cost saving of £26,000 and 151 tonnes less in carbon dioxide emissions from April 2009."

Perth and Kinross Council already had a set of servers that were coming to the end of their service life, and it thus was faced with the prospect of refilling the racks in the datacenter or else moving the OS installations into virtual machines. It had a hardware budget for a conventional upgrade of some £150,000, but managed to spend only one-third of that sum by opting for virtualisation. Then there were the savings accrued from reduced electricity consumption, and a consequent reduction in cooling costs too. And, at the end of the day, it got a virtualised cluster that also allows it to have a highly flexible disaster-recovery scheme, together with rapid deployment of test VMs as necessary for prototyping work.

I'm seriously impressed by what seems to have been achieved in this effort. It really is imperative these days - especially as we slide into a recession - that we get the most computing horsepower for our investment, and that means servers humming along with their System Idle process taking 99.9% of the processor cycles simply aren't good enough any more. Clearly, the benefit in your own machine room will be determined by the number of physical servers you have today, the loadings on them, and the facilities - in terms of cash to spend, new hardware, and people skills you have on- and off-site to help in making the change.