Google's Compute Engine 600K VMs = 37,500 servers & 15MW of data center power

I am probably wrong on this calculation, but willing to take a stab at what could be.

Google announced its Compute Engine offering this morning.

A Genome calculation with 10,000 cores was used as a demo.  If you use an 8 core Intel Xeon with dual processors this is 625 servers.

Urs Hoelzle was presenting that there are 771,886 cores ready to run a Google Compute Engine job.

Here is a the calculation executed with 600K cores with refreshes in seconds vs. 10 minutes for the 10,000 core system.

So how many servers would 600K cores be?  If you assume 8 core processors (I was going to assume 10 core processors, but backed off to 8).  The following is pricing info.

How many servers could this be?  600,000 Cores/16 cores/server = 37,500 servers.  (assume a virtual core does map to a physical core, note this is not always true)  Assume 350 Watts per server ( 2 processors, 64 GB more of RAM, 2 HD) is 13.125MW of IT power with a PUE of 1.12 you get to 15MW overall data center power consumption

600K cores may seem big.  But, thinking about 37,500 servers and 15MW of power is really impressive at least to the data center geeks.  Oh yeh, there 771,886 cores available which is 48,242 servers, 16.9 MW of IT and 18.9 MW of data center power.

It is impressive to think one whole 20MW of data center capacity is available on demand for a Google Compute Engine job.  Keep in mind this what is available, not the total capacity and being consumed.

Cloud Oxymoron: 91% IT decision makers like the cloud, yet 43% business users skip central IT

Rackspace has a press release on its survey that finds nine in ten IT decision makers support the cloud.

Rackspace Hosting Reports Nine in Ten IT Decision Makers in National Survey Have Positive View of Cloud Computing

0

Overwhelming Majority of Decision Makers Believe Customer Service and Technical Support

Are Important Considerations When Choosing Cloud Provider

San Antonio, TX, June 25, 2012 – Nine in ten (91%) of IT decision makers have a positive opinion of cloud computing according to a national survey commissioned by Rackspace Hosting.  IT decision makers view customer service and technical support as important considerations when choosing a cloud computing provider.

But, when I read the press release an Oxymoron comes out.

An oxymoron (plural oxymorons or oxymora) (from Greek ὀξύμωρον, "sharp dull") is a figure of speech that combines contradictory terms. Oxymorons appear in a variety of contexts, including inadvertent errors such as ground pilot and literary oxymorons crafted to reveal a paradox.

The contradictory idea is the same people who filled out the survey have a wide number of users that bypass the corporate IT group for cloud services.

Rogue IT is Prevalent

Nearly half (43%) of decision makers surveyed are aware of people taking it upon themselves to use cloud computing services or resources not provided by their organization’s IT department in order to help with work projects.   Among respondents reporting someone in their organization using cloud services independently of the IT department, 38% said the main reason is to save time.  One in three respondents believe it was because the solutions were not available internally or because it was a way to not deal with the organization’s IT department.

Now you can explain this many different ways.  Let me give you simple way to look at this.  Corporate IT is a huge group with lots of lots of employees.  The Cloud is a threat to their jobs.  So, whenever a cloud project is brought into corporate IT, a lot of what goes on are attempts to educate the users that they should be using corporate IT staff to run their cloud projects.  This delays the project and increase the costs.  

Wouldn't it be amazing if corporate IT said we can execute cloud projects faster and cheaper than you can on your own.  Shouldn't this be true if corporate IT really had Cloud skills?

What the cloud is doing is giving business users a freedom of choice.  The past corporate IT was a monopoly.

“Cloud computing is spurring innovation by enabling business users and developers to deploy, configure and adapt faster,” said John Engates, chief technology officer at Rackspace.  “The survey confirms the rapid migration to the cloud as IT leaders reap the benefit of spending more time creating and less time configuring.”

The harsh reality of the cloud is it eliminates jobs in corporate IT.  If it doesn't eliminate jobs how do you expect to save money?  Rackspace, Softlayer, and Amazon Web Services would go broke if they had the manpower overhead of corporate IT.

The evil part of The Cloud is it is threat to the large mass of people in IT.  Amazon, Facebook, Google, and others are proud they have one sys admin taking care of up to 10,000 servers.

Cloud API Fight at GigaOm Structure

One of the more entertaining sessions was on  Cloud APIs.

API WARS: DELIVERING THE DE FACTO STANDARD

 

OpenStack. CloudStack. Amazon now lets Eucalyptus customers link their private clouds to AWS. The cloud industry has grown up, and after six years, Amazon is still on top. Do the open-source efforts have a chance, or is this recent fragmentation the last straw?

Moderated by:Jo Maitland - Research Director, GigaOM Pro
Speakers:Sameer Dholakia - Group VP and GM, Cloud Platforms Group, Citrix

 

Chris C. Kemp - CEO, Nebula and Co-Founder, OpenStack

 

Marten Mickos - CEO, Eucalyptus Systems 

The video is here.

Watch live streaming video from gigaomstructure at livestream.com

If you don't want to watch the video here is a post on the presentation.

If AWS is the WalMart of cloud, is OpenStack the Soviet Union?

By Kevin Fitchard Jun. 21, 2012, 10:56am PT

Some of the most dynamic part of the presentation was this discussion..

Kemp took Citrix and Eucalyptus to task for reinforcing Amazon’s dominance rather than embracing the OpenStack project. As you can imagine, Eucalyptus Systems CEO Marten Mickos and Citrix Systems Cloud Platforms Group GM Sameer Dholakia took exception to Kemp’s claims, particularly his characterization of their cloud platforms as closed.

Both pointed out that their platforms are open-source, just like OpenStack, but Kemp refused to accept that definition, saying the companies developed the core of platform internally and then released their software to the open-source community. Kemp contrasted that with the OpenStack, which is developed top-to-bottom by its broad membership with no large company having any outsized influence.

Google's Story of Google Apps, 65-85% less energy, US GSA migrates 17,000 for a $285,000 savings (90% energy reduction)

Google had a viral video on The Story of Send.  Google just released a post that fits in the same theme with what could be "The Story of Google Apps."

The blog post is backed up with a paper.

The data that gives this story credibility is the win Google had for Google Apps with the US GSA.

Lower energy use results in less carbon pollution and more energy saved for organizations. That’s what happened at the U.S. General Services Administration (GSA), which recently switched its 17,000 users to Google Apps for Government. We found that the GSA was able to reduce server energy consumption by nearly 90% and carbon emissions by 85%. That means the GSA will save an estimated $285,000 annually on energy costs alone, a 93% cost reduction.

The specifics of the GSA story are here.

It wasn't too long ago that any small business that wanted e-mail was seriously looking at putting an e-mail server on site.  What small business does that now?  Along with that came Office licenses.  Google is targeting this whole Office document e-mail system with cloud services that are much more efficient.

A typical organization has a lot more servers than it needs—for backup, failures and spikes in demand for computing. Cloud-based service providers like Google aggregate demand across thousands of people, substantially increasing how much servers are utilized. And our data centers use equipment and software specially designed to minimize energy use. The cloud can do the same work much more efficiently than locally hosted servers.

All of this is built on Google's green data centers.

We’ve built efficient data centers all over the world, even designing them in ways that make the best use of the natural environment, and we continue working to improve their performance. We think using the super-efficient cloud to deliver services like Google Apps can be part of the solution towards a more energy efficient future.

Posted by Urs Hoelzle, Senior Vice President for Technical Infrastructure

(Cross-posted on the Google Green Blog)

Understanding Data Center Types - Cloud, Hosted, Colocation, Wholesale, Owned

I was having a conversation with a client and it occured that most company executives probably don't understand the different data center types that exist.  It seemed worthwhile to describe the different data center types that exist and how executives should understand the differences.  They hear terms like cloud, hosted, colo, and wholesale all the time, but what does this mean?

Besides doing a few searches, I reached out to Jones Lang Lasalle's Michael Siteman to see what he had on data center types. It was quite thorough.  I needed something simpler.  Something a non-data center executive could understand in one ppt slide.  

I am sure this seems obvious to most of you, but trying to get this into one slide was a good exercise.

Here is my current thinking a slide.

The text is here.

•Cloud (VM) – bring your code and data - OS, Server, Network, Storage available for lease; on-site operations and IT services all done by cloud provider

•Hosted (Servers) – Physical servers are unit of delivery within IT environment design – HW for lease

–Similar to an internal IT service group for small scale

–Cloud Hybrids are more common

•Colocation (Racks) – bring your IT equipment, pick your ISP, provide power, space, facility operations

–Lease space and power capacity – for example 10 racks @ 10KW/rack

•Wholesale (MW of capacity) – you rent space, power, cooling and an open floor plan; you decide the layout of your space – power, cooling, network

–Need facility operations as well as IT Operations for your space

–Lease 10,000 sq ft @ 1MW

•Owned DC (everything) – you have everything under your control

–Big things gained vs. the previous steps you pick the site, you pick the design best meets your business needs, and the whole facility is yours

–Problem: if you haven’t built a lot of DC, then you will make mistakes

–2MW plus, the big big guys Google, Facebook, Amazon and Microsoft are building 10 – 30MW

 

I think this works pretty well.  Technical enough with details like power consumption, yet high level to convey the differences.