Dr. Vishal Sikka, chief technology officer (CTO) of SAP, was one of the keynoters at the May 2009 Interop conference on cloud computing. With a couple of crisp statistics, Sikka cut to the heart of the matter as to where and how the cloud works and where it does not at least at this point in its maturity. (And as he said, maybe never completely, if you consider that we all still carry one or more batteries with us despite all the electricity available in the "electric utility cloud.")
Mr. Sikka overviewed the cases of a half dozen typical SAP customers in various industries that are working with terabytes of data in memory, handling billions of transactions a year, and running analytics against the combination of the two in real time. (Spend 30 minutes listening to his keynote online and you will immediately recognize who these large SAP customers are. They are all household brand names.)
He actually considers that some of these customers are already using cloud computing but not in the sense of an insular application like sales force automation running on a service provider's multi-tenant server farm being paid for by the drink. As I wrote on June 3 in another context, I agree. It's just that 30 years ago it was called SNA or DECnet.
The SAP CTO said a key to what you can do in the cloud is to apply the same analysis to your cloud computing plans that SAP applies to all its product development efforts for which he is responsible. He asks the questions, does the functionality support:
- Breadth: in terms of geography as well as size of company (the company using SAP and the companies the SAP customer deals with)
- Depth: not just transaction processing (TP) but analytics and business intelligence (BI)
- Longevity: consistently and somewhat seamlessly for generations (he made the point that SAP has 100 customers-out of its 86,000-that have been doing business with SAP for over 30 years). The concept of "generations" applies not only to application generations but the changes in underlying middleware and systems/storage technology.
By this analysis, the cloud as it is commonly defined has a long way to go. No doubt it provides scale and ease of consumption but it still lacks a lot in terms of integration support (most clouds are islands), integrity (most clouds suffer notorious downtime and some have even gone out of business) and depth (all have issues with complex analytics, managing complex orders, and especially putting those two functions together in things like availability-to-promise processing). That's why although I often make the point that the cloud is just the natural evolution from timesharing I also say there is something to be said for evolution.
You many not agree with Mr. Sikka's conclusion but it makes sense to use his three criteria or something like them of your choosing before sailing off into the ether with your organization's key asset - its information.