Well, OK it doesn't change "Everything", but it will have a significant impact on your strategic and tactical infrastructure plans from your Data Center to the server.

The promise of Cloud computing is many fold, but I don't think that promise is understood very well as it relates to your infrastructure & facility design. 

Today we build data centers based on a "Tier" model. Tier I is basic connectivity, power & HVAC with no redundancy. A Tier IV data center has full fault tolerance and redundancy built in to almost every aspect of the facility.

A Tier I facility can expect around 99.5 percent availability at a cost of $500 - $600 a SF to build (depending on power density and cost of land etc.)

A Tier IV facility can expect 99.999 percent availability at a cost of $2200-$2500+ per SF to build. There are also higher run rate costs associated with facility maintenance and power use as a result of all the redundant power and cooling equipment.

A 10K SF Tier I data center would cost roughly 6 million. A 10K SF Tier IV would cost roughly 22 million. In either case you would also need to account for at least a minimal data center failover capacity.

In a perfect "Cloud" world the following would be true:

·      Your compute load/applications will be everywhere and nowhere.

·      Your compute, storage and network resources will be measured on capacity and each component will allow for failure in place. In other words, no one component will bring the system down.

·      You'll only use the resources you need, when you need them and it won't matter where those resources come from or who the manufacturer was.

·      Your data center will talk to your infrastructure and warn it of impending doom and the infrastructure will react to that warning by moving work load to a safe location.

·      Application availability and performance will be improved no matter where you're accessing from.

·      Reduced staffing requirements, but higher level talent.

·      etc., etc..

There are other things that make the perfect "Cloud", but the above are certainly key areas of opportunity.

oNormal" style="MARGIN: 0in 0in 0pt; mso-margin-top-alt: auto; mso-margin-bottom-alt: auto">So, assuming the perfect cloud comes to fruition and I believe it will based on the number and quality of vendors pursuing it, there will be considerable change in your infrastructure. Don't worry, it's good change! The Tier III & Tier IV data centers of today are dead in the cloudy world of the near future. The following are some savings opportunities associated with having a "perfect" cloud environment:

·          Instead of a Tier III/IV data center with a Tier I/II failover site you'll be able to build two sites for less the 30% of the cost and each site will be active. This means you can build in your geographic diversity (disaster avoidance) for less than it would have cost you to build just one Tier III/IV data center. Many large companies these days are spending $100 million or more on their newest mega data center. Imagine if you could provide 2 active facilities for less than $50 million. You'll also improve your site efficiency (better PUE) and be saving year over year on mechanical & electrical maintenance. Or you can avoid building a data center altogether and make hosting companies fight over your business.

·          You'll be able to buy the server, storage and network equipment that has the best price at the time of need. In the perfect cloud you won't need to worry about whether you're buying the latest HP or a "white" box provider's equipment, because each box is merely capacity, not availability.

·          You won't need redundancy in the servers themselves. Today the vast majority of us buy servers with 2 or more power supplies, 2 or more network interface cards and remote access cards. Not only do the extra power supplies add $200 or more per server, they also reduce your power use efficiency by several percentage points. This area of opportunity alone can mean between $300 & $600 per server in savings along with a small amount of run rate savings over the life of the system. You also won't need to buy high priced service and support contracts anymore.

o        Example: If you buy 1000 servers per year your savings would be roughly $450K per year:

·         $400 per server, plus $50 a year through power efficiency

·          Human resources costs will go down because you’ll be able to run in lights out mode. You should be able to schedule server add/replacement on a weekly basis because your equipment will be fail in place. Running a lights out operation will also further reduce your power use and other costs associated with supporting humans on site.

There are other opportunities to save in your infrastructure, but the above should at least wet your appetite for the "perfect" cloud.

Cloud presents a dramatic improvement opportunity in the availability and performance of your applications. The underlying infrastructure will be more effectively utilized and will require considerably less redundancy and support. A positive side effect will be that you're helping to save the planet because you'll be using less power.