Microsoft.com published a report on their results running virtualized MSDN and TechNet Hyper-V servers.  Finally, someone steps up and says how much the overhead is for virtualization layer, 5 - 6%.

Hardware

Dual socket Quad-Core Intel processors
32GB RAM
4x146GB disk drives

Virtual machines

4 Virtual processors
10GB RAM
250GB dynamic VHD

Operating system – Parent

Windows Server 2008 Hyper-V RC0
Reserved 2GB RAM from 32GB total

Operating system – VMs

Windows Server 2008
Internet Information Services (IIS) 7.0

Performance #'s

· Hyper-V CPU overhead (as measured by the parent partition utilization) was 5 to 6 percent with linear progression as the number of requests increased. This is illustrated in Figure 2.

· CPU oversubscription (three four-processor VMs on an eight-processor physical server) resulted in 3 percent lower overall performance per physical server based on overall requests per second per 1 percent CPU.

· Requests per second per 1 percent CPU performance of MSDN over the previous physical server platform improved. This demonstrates to us the viability of efficient consolidation from dedicated older physical servers to shared virtualized platforms.

· Physical MSDN handled 21 percent more requests per second per 1 percent CPU than virtualized MSDN. Figure 3 illustrates physical compared to virtual scale up for MSDN under peak production load on matching hardware.

Lessons Learned

· Based on our results with Hyper-V, and a review of current hardware utilization across our environment, we expect to realize significant benefits by consolidating diverse applications on a virtualized platform. Mixing high and low scale systems on the same physical server resources should enable us achieve improved overall hardware utilization and a reduced physical footprint.

· Although we reduced our hardware supporting MSDN and TechNet by migrating from older physical servers to a smaller deployment of new, more powerful physical servers, there is some minimal overhead associated with virtualization. From our perspective, Hyper-V has clearly delivered enough performance, stability, and scale to drive widespread adoption in our production environment. The flexibility and management gains expected from coupling System Center Virtual Machine Manager (SCVMM) v2 with Hyper-V should justify the measured overhead.

· If the three percent or less in additional overhead from oversubscribing resources is consistent with additional application workloads tested, oversubscription with Hyper-V should provide MSCOM Ops significant flexibility and reasonable performance for application consolidation.

· The results we achieved are based on the application characteristics of MSDN and TechNet on Hyper-V RC0. We are working to build a model based on this data that we hope will allow us to predict physical and VM requirements based on common performance characteristics of a Web site. The model will likely include the current requests per second per 1 percent CPU metric with additional memory and I/O performance characteristics as we virtualize other Web applications and gather more test and production data.

WSJ article about Ann Livermore, head of HP's storage and servers, software and services business.

She also is focused on customers, talking to two or three big ones every day. And she has a reputation for identifying market trends. When she visited a dozen financial-services companies several years ago, she listened when IT executives said they were spending too much time and money operating their data networks.

Out of those conversations came H-P's strategy for building "the next generation of [automated] data centers, which is now driving growth in software and services," says Ms. Livermore.

And, also credits her for working on the EDS deal.

Ms. Livermore helped develop strategy for the EDS deal with Mr. Hurd. Like him, she knew H-P on its own couldn't expand its outsourcing services enough to meet demand.

News.com has its Google data center report. And, as Data Center Knowledge points out.

In each cluster's first year, it's typical that 1,000 individual machine failures will occur; thousands of hard drive failures will occur; one power distribution unit will fail, bringing down 500 to 1,000 machines for about 6 hours; 20 racks will fail, each time causing 40 to 80 machines to vanish from the network; 5 racks will "go wonky," with half their network packets missing in action; and the cluster will have to be rewired once, affecting 5 percent of the machines at any given moment over a 2-day span, Dean said. And there's about a 50 percent chance that the cluster will overheat, taking down most of the servers in less than 5 minutes and taking 1 to 2 days to recover.

A 50 percent chance that the cluster will overheat? This suggests that Google's approach, which packs 40 servers into each rack, is running pretty close to the edge in terms of thermal management. Or perhaps that Google has trouble anticipating when an area of its data center may develop cooling challenges.

And, as I wrote a post about gigaom.com, Nicholas Carr's posted as well on Understanding Amazon Web services.

There are two ways to look at Amazon.com: as a retailer, and as a software company that runs a retailing application. Both are accurate, and in combination they explain why Amazon, rather than a traditional computer company, has become the most successful early mover in supplying computing as a utility service. For Amazon, running a cloud computing service is core to its business in a way that it isn't for, say, IBM, Sun, or HP.

Now, if you were a customer who would you buy web services from?  Amazon or Google. 

This week's Barron's writes a favorable review of Eaton, discussing its diversification into products for data centers and investing in breakthroughs in conserving energy.

Eaton's trucking business, meanwhile, is increasingly focused on energy efficiency. Just recently the company (ticker: ETN) won orders for 207 hybrid diesel-electric transit buses for Guangzhou, China, and 200 hybrid-electric vehicles for United Parcel Service (UPS).

"There's a real economic opportunity for companies doing progressive innovations to identify technologies which will allow breakthroughs in conserving energy," Cutler tells Barron's. "For Eaton, it really is a sweet spot of combining our customers' needs for more sustainable business along with our own base set of values."

And, for those of you who buy Eaton's data center products.

Eaton's new Powerware products for uninterrupted power-supply systems, which help lower energy costs of data centers, contributed to a more than 30% increase in sales last year to Russia and the Commonwealth of Independent States alone.

They are so green, they are being added to an environmental sustainability mutual fund.

Eaton's green leanings have prompted Portfolio 21, a global mutual fund based on environmental sustainability, to start building a position in the company.

If you want to see Eaton's green product and service commitment, check out http://www.eaton.com/EatonCom/Markets/Electrical/Sustainability/EatonsGreenProductsandServices/index.htm

Eaton is one company who can proudly say they have bet on Green successfully and are performing well in this economy.

GigaOm has an interview Amazon's Jeff Bezos about Amazon Web Services.

This video confirms 2 reasons why I think AWS is successful. Being a retailer, Amazon has low margins and needs to be efficient. (they know how to do pricing to maximize profit.) And, being a retailer, they have to hit dates to meet retail schedules (no compromises on delivery).

Here are more details for you to digest.

• How and when Amazon began its cloud computing effort.
• Why Amazon has become an innovator with Amazon Web Services and how it relates to their core business of being an online retailer.
• Whether or not Wall Street recognizes Amazon’s cloud efforts.
• What’s next for Amazon Web Services.
• Whether or not Amazon has plans for a VC fund or for cloud computing startups.

For even more info about Amazon’s cloud computing efforts, join us at our upcoming conference, Structure ‘08, where CTO Werner Vogels will be delivering a keynote address.