Four Great Ways to Fail at SOA

Loraine Lawson

Business in general and IT in particular are often obsessed with finding "best practices." I suppose the logic is, if you know what constitutes a best practice, you can duplicate it and create your own success.


But the reality is, mistakes can often teach us just as much, if not more, than those best practices-assuming we take the time to actually learn from them. Just ask any successful entrepreneur.


So, in the spirit of learning from our mistakes, I've found a little weekend reading for you: "Four SOA Implementation Worst Practices: Why Service-Oriented Architectures Fail and How to Prevent It."


This article, published on Information Management, is by Jake Freivald, the vice president of marketing at business intelligence and Web reporting software provider Information Builders. Now, I know what you're thinking. This is going to be another list that tells you SOA's failure is because you forgot governance or something like that. While that is a commonly cited reason for SOA failure, I think you'll find Freivald's list offers a fresh look at SOA mistakes. Here it is:


  • Overemphasizing low-level code.
  • Centralizing design and development.
  • Ripping and replacing legacy software.
  • And buying software without support.


See what I mean?


Freivald is specifically tackling worst practices related to using SOA as an integration solution. Now, I know, some people say SOA shouldn't be used for integration. The whole point is to avoid that point-to-point, tight coupling stuff. But if it's done right, SOA can solve a lot of your integration problems. So, let's move on ....


This is also a great article for those who are struggling to define services, because many of these problems boil down to how you deploy services. For instance, low-level code is all about making services too small in impact. So the article spends a good deal of space discussing what a service should be, what it shouldn't be, what happens when services are too granular and so on.


He also tackles another common misstep: Mistaking SOA for enterprise application integration, v. 2.0.


Another interesting item in this insightful article is Freivald's observation that these four mistakes are typically "the result of wanting to ride the latest technology wave without balancing the hype with practical knowledge and experience." And really, isn't that the basis of all too many IT mistakes?


After you've finished reading about the worst SOA practices, balance your reading by checking out these posts on what it takes to ensure SOA success:


Have a good weekend.

Add Comment      Leave a comment on this blog post
Mar 12, 2009 11:27 AM Jignesh Shah Jignesh Shah  says:

It's interesting that "Centralizing design and development" of services is listed as worst practice. We increasingly see customers employ a "Shared Services" organization to their fledgling SOA proper direction and momentum. Such an organization is tasked with identifying, designing and delivering services around the core systems that have an enterprise scope. There are several advantages including being able to look holistically across the enterprise in shaping the services portfolio, simplifying the funding model and cutting through political barriers.

Freivald is correct that such an org would have to be well versed in systems they building services on top of. Which is why, the integration team is often the defacto choice for SOA shared services - this group has an enterprise perspective, has experience dealing with core systems, understands the challenges of heterogeneity and has the required technical skills in terms of middleware.

Freivald's suggested "Decentralizing Service Creation" approach would practically require a Utopian company where interests of others are placed before the interest of each group. It simply does not happen for most large organizations getting started with SOA.

Jun 16, 2009 3:43 AM Randy Williamson Randy Williamson  says:


There is a fifth way ... stick your head in the sand and hope it goes away. There are still those in IT management that consider SOA as "unproven, pie in the sky technology".

With backward thinkers like this in charge of the direction of once proud, award winning leading edge organizations, we can see thier decline into the IT neverworld of laggards instead of creating the future.


Post a comment





(Maximum characters: 1200). You have 1200 characters left.



Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.