We have come a long way since we launched the Sun Grid Compute Utility last year.  Sun Grid when launched was the world's first true compute utility made available over the network.  It is a compute utility that provides users with access to enormous compute resources to run any sort of complex calculations.  The Sun Grid basically provides "infrastructure on-demand," where users bring their own applications and data, upload them and run their projects.

Since its launch, we have seen a number of customers utilizing the grid for various things. I have blogged about some of them in the past, such as CD02 that uses the grid to conduct risk analysis for its customers, and AMD that is tapping into the grid to simulate processor performance. The case of CD02 is interesting as access to Sun Grid allowed them to grow their business and service more customers, something that would have been hard if they had had to build all the IT infrastructure in-house.  I could also mention Applied Biosystems who performed compute-intensive data research on Sun Grid to develop millions of new genomic assays. 

These, and all our other customer, have benefited from the utility model: none of them had to make large investments in IT infrastructure; none of them had to sign a long term contract locking them in for an extended period; all of them were able to shorten the time to results; and all of them only paid for the compute power they used.  Such is the nature of a true utility offering.  You come and you go as you please.  You use what you need, and you only pay for what you use. If someone wants to offer you a "utility," check if it meets this criteria.  If not, they are probably trying to sell you on traditional hosting.

Another benefit of using Sun Grid Compute Utility is that users do not need to make large capital investments in IT infrastructure, not to mention worry about operation, space, power, and cooling issues. That's Sun's problem, not yours.  IT becomes an operational expense, not a capital expense. 

The launch of Sun Grid has been a tremendous learning experience for us.  From every perspective.  There are of-course significant engineering challenges associated with building a multi-tenant, secure, and rock-solid public compute utility service that is offered over the Internet.  But there are also a lot of interesting business and legal challenges we have encountered.  The pricing model of $1/CPU-hour with no additional cost or hidden fees sounds simple enough for customers. That's what matters in a utility model: Simple for the end users, and all that complexity hidden from view.  After all, when I plug my laptop in the wall to charge the battery, I don't think about how the electricity travels from the Hoover Dam (or wherever it's generated) through the PG&E electrical grid to my house.  The same goes for the compute power you get via the Sun Grid.

To give you some insight into the legal issues we have had to resolve, you may want to check out the blog that Matt Dillon, Sun's General Counsel, wrote earlier on Sun Grid:

"The idea behind the Sun Grid is that in the future, just as with electric power, individuals will access their compute power, software applications and storage over the Internet. It is an intriguing idea that presents an interesting, but complex array of legal and regulatory challenges. In the course of developing the Sun Grid we have had to consider legal issues involving: data privacy, data security, regulations applicable to on-line commerce, DMCA, tax, Sarbanes-Oxley, Digital Rights Management, state and federal consumer protection statutes, licensing of third-party applications in a hosted environment and US export controls."

I like Matt's analogy on comparing compute power with electricity. We have done that before, see Jonathan's blog.  Nick Carr, more than most others, has exploited this point.  While I may not agree with all his statements, I believe he hits the nail on the head when he talks about the future of computing being utility.  As a side item, I should also mention his "Trailer park computing" blog he wrote on Sun's Blackbox, a mobile datacenter in a container, where he argues that it is one more manifestation of the fundamental shift in computing.

Carr's take on what he believes to be Sun's approach to utility computing is also interesting:  "It's the Thomas Edison strategy. In 1888, Edison built the world's first electric utility - the Pearl Street Station in New York City. It was a marvel of engineering, but it took a couple of years before it turned a profit. Edison didn't care, though. He built the station as a model for others. And his plan worked. Pearl Street set off a boom in the construction of similar generating plants around the country, some operated as public utilities, others operated privately within factories. Edison's company - General Electric - made its money by outfitting those plants, not by running them.  My guess is that that's what Sun is hoping to do as well."

I think any reasonable person with knowledge of the IT industry would agree that utility computing is the way of the future. The question is really how and how fast will it be adopted.  I think this will be an evolution; customers are not throwing out their IT infrastructure to go all online.  But as new IT needs emerge, they will increasingly look to online services to meet that need.  Right now for some projects it makes sense, for others customers will opt for running the systems in-house.  It will adopted at different rate by different industries and for different types of projects.  Overtime the pendulum will swing towards utility computing, and when it swings, Sun with it's utility computing expertise, will be in a position to lead the way.

So what's next?  Well, first we had to build out a rock-solid, secure infrastructure.  Now it's time to take it to the next level.