jan.m, I haven't seen discussions about new technologies that can overtake BEAS. Please do point me to them. I am curious to see the reports.
Generally speaking, applications (B2B/B2C/Enterprise/small or big) are built over certain layers of software.
The bottom most layer is the Operating System (UNIX/Windows NT/MVS/..), which is most easily understood.
The next layer is the distributed transaction framework (what CICS was on mainframes). This is a layer that is becoming more popular, as applications want to scale.
From the beginning of distributed computing, there were higher level tools (used to call them RAD tools/packages) that helped build apps faster, where there are more application-friendly abstractions (or utilities). (Example: Forte).
In the e-commerce age, Vignette is one good example. It does provide a framework, with a number of higher level abstraction facilities, that help build apps faster. These apps are reliable and scalable too.
Typically these types of RAD tools can take you far, but not all the way. It is the next big step, beyond building a site using MS Front Page.
But you cannot build a B2B apps with these.
Then come the specialized RAD tools for larger apps, such as Ariba. Ariba would definitely have more octane to build much bigger apps. But they also would be inventing a new transaction framework (a la WebLogic Enterprise) in this process. We know how long Ariba has been around, at least 4 years. They do have strong R&D. I am sure their current framework works for some apps, better than WLE in some aspects, but would still be off to be a full-fledged generic framework.
Ideally, Ariba-like companies should use BEAS WLE to take full advantage of a proven framework. Today, any other company can start a Ariba like product, by simply investing in BEAS WLE. That is because, Ariba must've spent more time in building the framework than on their core product.
20 years ago, technologists would think of even rewriting the OS - if they didn't 'like' it. 10 years ago, they would question a databse package. Why not build a RDBMS/a network modelled DBMS .. by ourselves, instead of buying Oracle? Now, considering the maturity of OSs and RDBMSs, no one questions buying an off-the-shelf product.
The same level of comfort has not yet been established - in the field of 'distributed transaction frameworks'. All it takes is another company to start competing the likes of Ariba, with the help of established frameworks such as WLE.
Looking forward to the links.
- Zelix |
