Shai Agassi, executive board member at SAP - one of the largest software companies in the world - was the second keynote at Accelerating Change Conference 2004. He's responsible for SAP's overall technology strategy. [Agassi was much more engaging and humorous than my notes indicate.]
I am considered the futurist in our company. Our company is very sales-driven so we're making futurist predictions when they talk about the next quarter - but I myself look out five years.
Nicholas Carr asks in his book, Does I.T. Matter? He equates computer industry to the train industry after all the train tracks have been laid. But the computer industry is different. Most of I.T. investment happened in the last 20 years reduces process execution time. We're at 200 millisecond transactions. So the question is valid: Why would you invest to go from 200 millisecond to 20 millisecond transactions?
The next challenge is "time to change." Product lifecycles haven't moved as fast. And change management - pushing beyond the corporate inertia - events such as a merger takes 18 months in a corporation. A corporate venture can take 18 months to reflect an effect in the business.
We've look at systems that define core and context. (You have to read two books by Geoffrey Moore to work in the Valley ;-))
|focus: differentiation||focus: productivity|
|Mission critical advantage||Innovation||Standardization|
Source: Geoffrey Moore's book, Living on the Fault Line.
For an airline, a core process is to fly the plane. "We bring most of our airplanes back down to earth" is not a differentiator.
It's circular: starting at Invention -> Innovation -> Standardization -> Commoditization.
We're starting to see emergence of something we call composition. There is a company in Europe that looks at the PC supply chain and asks every Monday what's cheap? A bit like using Google to type in what's in your fridge and getting a recipe back. And then they offer that for sale.
New I.T. capabilities required for this composition evolution. What we call applistructure - the merging of applications and infrastructure.
The four players are Microsoft, Oracle, IBM and us. Our stuff makes your company run but doesn't really differentiate it. IBM says "we'll run the stuff that's not critical to your business and we'll do it on demand". With Oracle we'll differentiate and put it in your database - but it's one-off. Microsoft is often where most people start with but it's arguable whether it scales.
(Agassi writes in Microsoft in the Invention quadrant, Oracle in Innovation, IBM in Commoditization, and SAP in the Standardization quadrant in the Core vs Context quadrant above.)
There's about 10,000 objects that are like the language or lingua franca of business - it's the core stuff that's in most of the companies in the world. Now at the layer above the composition platfrom (SAP's NetWeaver) will be about 10,000 companies. [Hmmm, SAP has never been a platform play before. This will be interesting.] Composite applications are built using this platform.
Three tiered software - database, application and client. Oracle standardized the database and Microsoft the client. In the middle there were many many applications - JD Edwards, SAP, etc. Composite solutions are what will be bought and sold. You don't buy the parts for a car, you buy a ready-made car. This vision will take 5-8 years.
Linux is attacking at the desktop. And database engines will get attacked by companies like mySQL.
Longer Term Trends
|services platform||database||file system|
|devices ||clerks||not knowing|
|domain specific language||java||cobol|
|autonomous systems ||system integration||IT management|
1 - including, RFID enabled objects
2 - desktop become secondary
3 - cheaper people aren't solution; we're moving away from code to visual models of process flows that can be managed by business people; this is as big a shift of what happened in the telco industry ...at one time they were worried about running out of switchboard operators; code is hard to maintain. You'll see more system design oriented people than coders - changes the skillsets not necessarily number of people employed. And you need to be close to the user to create these models.)
4 - system monitors its own health and its own usage