Cloud computing is not just a name for outsourcing
- 11 April, 2011 23:30
I'm guessing many of you are asking if cloud computing isn't just a new name for ASPs, software as a service, outsourcing or, for us older guys, timesharing. While the cloud certainly shares principles with all of them, something more significant is happening, something with the same impact as the generational shift from mainframes to client/server.
Let me explain just how significant it is.
In 2008 I launched a lecture series on cloud computing at Beijing's Tsinghua University. Amazon was kind enough to donate $3,000 worth of compute time to my students. On the first day of class, I told them $3,000 could buy a server in Ireland, Virginia or California for 3.5 years. They yawned. Then I told them that same $3,000 could buy 10,000 servers for 30 minutes. They stopped yawning. No one has ever been able to do that before. Of course, what would one do with that many servers? Well, people are already using thousands of servers to do tumor diagnostics, risk analysis and load testing. Perhaps one day this massive capacity will be used to deliver more down-to-earth services, delivering relevant information whether I'm a healthcare, retail or home-improvement customer.
What else lies ahead? On the application front, there are plenty yet to be built. Some may emerge from your organization, just as Sabre came out of American Airlines. As for compute and storage, there's every reason to believe there will be cloud services specialized by location, security or performance characteristics. Today there are a handful of compute and storage offerings, but look at any hardware vendor's catalog and you will see hundreds of different kinds of servers and storage. Why wouldn't this be the case in the cloud? Today's compute and storage cloud services are located in a small number of countries. But Canada, Mexico, Brazil, India and China will get in the game too. And while every cloud provider is extending its security capabilities, expect enhanced services, such as two-factor authentication, for some special ones.
Given all this change, what should a CIO do?
Get smart, get specialized.
Create an office of cloud computing. Dedicate specialists to data center, compute and storage, platform and application cloud services. Have your application cloud service specialist inventory all your enterprise apps and ask the vendor to manage them.
Get smart about services provided by both public and private companies and use these in any new deployments. And not just business applications, but also the platforms that cloud services use to manage the security, availability and performance of your portfolio. Does it make sense to run spam-filtering software when you can purchase it as a service?
For your data center cloud service specialists, have them get smart on the cost of power, both in dollars and carbon. Have your compute and storage cloud specialists get smart on the existing services and business models. Get them to explain spot pricing to you and why you should do it. If it makes sense, consider building your own cloud service. Price it at X cents per hour and sell it to others in your industry or geography. Finally, get smart about search. I know you have SQL gurus, but who is your search guru?
For those of you who lived through the shift from mainframe to client/server, use the lessons you learned then to guide your thinking now.
That's right. We're getting ready to do it all over again.
Timothy Chou teaches cloud computing at Stanford University. He is the former president of Oracle On Demand and the author of Cloud: Seven Clear Business Models.
Read more about cloud computing in CIO's Cloud Computing Drilldown.