<< Cat   |   ARCHIVES   |   Osama Update and History >>

Giving Up on Enterprise-Wide Integration? Good!!

Giving Up on Enterprise-Wide Integration by Erika Morphy of newsfactor.com via Yahoo News

"Point-to-point integration is an excuse many companies use for their broader IT strategies," he added. "Companies need to wake up and smell the EAI coffee; all those hand-built connectors cannot ever scale, over time, the way an application layer can."

That from Lois Columbus at AMR research.

I used to be a big fan of fully thought out integrated solutions. They make "more sense" from a logical point of view, they are more scalable. They are more flexible. All those sorts of wonderful things. But after spending seven years at a company in the top 50 largest in the contry, more than ever I am convinced this is a counterproductive approach except in unusual situations.

The thing about integrated approaches is that to pay off you need time and money to do them right, and then they need time to grow and take root and start to bear fruit. This almost never happens.

By the time an integrated approach can be planned and implemented, the needs and priorities of the organization often have morphed unrecognizably. In the process of being built the integrated solution will have been compromized in order to meet time and budget goals. It will not be flexible enough to change to the new environment. Not to mention the fact that technology will have moved forward, already making the solution obsolete.

Now, perhaps some of this perception is specific to where I work and to my specific experiences over the past few years. Probably. Many people would probably tell me I am dead wrong and their experiences are completely opposite. OK. But I suspect that many of the underlying factors are widespread. There are certainly cases where large scale enterprise wide technical solutions are needed, and there is no other choice. In many more cases I'm convinced that need is just a phantom need. It sells software for the big vendors of integrated systems. Nothing more.

Yes, the spaghetti-like nature of just patching together systems as needed and making short term tactical projects to meet the immediate need done does not usually scale well, or provide a long lasting foundation that you can build on for decades.

But guess what? In most cases, it DOES NOT NEED TO. You solve the specific problem at hand, and move on. Yes, some of these "temporary" solutions end up living forever, but if they work, who cares? In most cases, the whole system will end up being revamped in favor of new technologies within a few years anyway, and/or the business needs for even the existance of the project at all will evaoporate.

So while the "Enterprise" quality solution would still be poking along trying to be planned and built, with ruggedness to survive for longer than it will ever be useful, the "quick and dirty" solution could already be operational and providing results now.

So yes... if you have a very stable organization, working on very stable systems and processes, and have a very long time horizon, and expect what is being built to last a very long time, and won't mind operating on obsolete technology five years on, and have lots of time and money to burn... go ahead with the big fully integrated solutions.

But if, like most of us, you need to get results quickly, and live in a world where business needs and priorities change at least once a year, if not more frequently, then beware of the big solutions. They will suck time and money and effort and most likely result in less than what you wanted. If you can get a quick and dirty that isn't TOO dirty, then go for it. You'll get the results you need faster.

And yes, EXPECT to replace it every couple of years with something new and EXPECT to spend some time troubleshooting and figuring out the spaghetti later. But guess what, you'll probably still end up spending less than the fully integrated mega-solutions.

The area where I've seen this the most is in content management, which has been my area for the last seven years or so. Perhaps I'll have some thoughts specific to that area later.


Abulsme - Wed, 30 Jul 2003, 10:46:06 PDT

COMMENTS



Note: You must be registered and logged in to post a comment!
If you try to post a comment without logging on first, the comment will be lost!
Apologies for the inconvienance. Blame the comment spammers.





Notify me when someone replies to this post?


Powered by pMachine