DMR: It seems then that the architecture and design and processes are much more important that the actual IT. 

Lew Tucker: I’ve always been a believer that architecture matters. You can have great processes and a weak architecture, and it’ll leave you vulnerable to security and reliability issues, among others. It always comes down to architecture. One of the advantages of Cloud computing is that it allows you to experiment more quickly with different architectures, because you can quickly bring up 50 virtual servers to explore how you might partition a database in a particular way, run it for a couple hours, and do some tests. When people ask how do I get started, I keep saying, “Just try it.” The cost of doing your own experiments is so low there just isn't much reason not to. If you’re just looking at the equipment costs, if I know precisely the demand, and it never changes, it’s almost always cheaper for me to buy it myself. But we’re forgetting the operational costs, and the opportunity costs that you give up because you are spending all this time maintaining your system.

DMR: Is there something missing where you say, “Gosh, I wish we would have that!”? And who would be a potential provider of that – besides Sun, obviously? 

Lew Tucker: I think that we already touched on a number of these things. An agreed-upon standard for Infrastructure-as-a-Service is the big missing element today. Secondly, we are recognizing that security is a major issue and that storage of personal information has associated privacy regulations that may differ for each country. Knowing where storage is physically located therefore becomes important. This may require us to always associate location metadata with each stored record. On the enterprise side, companies are looking for better Service Level Agreements and I expect to see competition between the different service providers around what kind of SLAs they will offer. 

DMR: Very interesting. What’s one of the exciting applications or technologies you see –   one that you can talk about? 

Lew Tucker: One area where I expect to see some radical changes is how the use of dynamically created virtual machines will cause us to change the way we design applications. With this new model, we will start to built applications as systems that are in some sense, aware of their own resource utilization, and can take steps to either increase the number of servers in response to increased demand, or decrease their resources to save money. Instead of having to file a trouble ticket with a system operator, saying “Please spin up four more servers for me,” now the application itself can do that provisioning through APIs. Of course, guardrails need to be put into place both internally and with the service provider because you don’t want that application to go and run up a $50 million bill in a week.