IBM walks the SOA walk

IBM walks the SOA walk

Summary: Over the past decade, IBM has trimmed down its own internal portfolio from 16,000 applications to 4,000, thanks to standardization and service-enablement.


As we’ve seen in recent months, IBM has been quite uber-eager to deliver SOA to the enterprise world, with countless product and service announcements. Big Blue talks the talk, but does it walk the walk? As a matter of fact, IBM has been service-orienting itself, and in a big way. 

Recently, I was given a glimpse inside IBM’s own SOA initiatives, courtesy of Howie Miller, IBM’s vice president for enterprise architecture.

Miller told me that Big Blue currently has 77 shareable and reusable services in production -- ranging from authentication to order fulfillment -- as part of its service-oriented architecture. The most notable outcome of IBM’s internal SOA efforts is the fact that Big Blue was able to reduce its inventory of 16,000 applications in 1998 to 4,000 applications today.

The secret sauce to streamlining down to a quarter of its applications was governance, Miller explained. “The governance model helped us achieve that nice statistic. Over the past decade, IBM has trimmed down its own internal portfolio from 16,000 applications to 4,000. It’s how we plan to go from more big integrated spaghetti code applications to more component-based Web services. That same governance model will be part of what gets us moving forward to the SOA-based approach.” Decisions about new technology investments are vetted through an Enterprise Architecture Council and Investment Review Board run by the CIO (Miller's boss). 

IBM’s CIO's office is leading the SOA charge internally at IBM, but doesn’t have a central bucket of money or huge staff to plan and build SOA. Instead, Miller, said, the office -- which has 10 architects -- plays a role in influencing Big Blue’s annual internal $4 billion annual technology spend. "About 28 percent of our investment of our total IT spend is what we call 'transformational investment,' or new development.” That's where SOA comes in.

“What we do, through roadmaps, standards, and target architecture is try to influence the behavior of the way that our development teams are doing transformation initiatives to spend their money.”

Miller described three of the 77 services IBM now maintains:

Web identity service: This is an authentication and authorization service for end-users accessing IBM applications. "When an employee or partner signs into a Web site, they have a single user ID and password for sign-on," Miller explained. "It used to be, every time I used a different application in IBM, I had a different user ID and password. Now, all these applications call the Web service to validate the user ID and password. This common Web service is callable from any application.”

Factory in a box: IBM's semiconductor group built a set of services called "semiconductor factory in a box.," Miller said. "It has enabled us to do is invest, acquire, or divest, or outsource manufacturing plant very quickly. If we divest a manufacturing plant, we had to rip out all the connections to various systems. Or, to acquire a new company that comes with a manufacturing plant, we had to integrate them into our systems. The process used to take several weeks -- now it's two hours.”

Customer order and tracking system: IBM's customer order analysis and tracking system (COATS) "was a big spaghetti-code application, all integrated into one big app," said Miller. The huge application was being overwhelmed by order volumes, which "were causing problems with performance with the application." MIller's staff decomposed the entire application into 13 separate components. "We found that the logjam in the application was only in a couple of the pieces of the app. We pulled out that application code for those two pieces and wrapped them as Web services. It was a way to rewrite a legacy application in pieces, without having to go do the full 13 big chunks of functional capability."

Miller says SOA enablement didn't happen overnight, even for IBM. "A lot of time, when we talk to folks about this stuff, they say, ‘oh my gosh, I could never do an enterprise architecture of that size, with the complexity of my company. It must have taken you years.’ The answer is yes, it did take years," he related. "But you can start in a corner of the business, business unit or functional area, and work it out from there."

Topic: IBM

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Seppaku by any other name

    If IBM can save money going with general web apps, why not everyone else? And if web apps are "general" then why pay big bucks to IBM? Web apps need to run "somewhere" (i.e. IBM hardware?), but they can run on Linux "commodity" servers, so where is IBM's AIX bait-and-switch?

    Question is - will IBM JUST slice across the abdomen, or finish the job slicing upward?
    Roger Ramjet
  • Re: IBM walks the SOA walk

    Nice fluff piece ... if you're into marketing.

    How about doing an indepth piece on the actual inhouse tools used by IBMers on a daily basis in teh performance of their 'value-adding' tasks and activities.

    When was the last time you used VM tools to access databases or other tools on a daily basis ... while working in an MS environment?

    How many times have you had to become a part of a beta-testing group where the latest IM, CRM or PMR tools are developed and tested internally before they're foisted on external customers at a hefty price ... all while being expected to meet corporate productivity targets ?

    Documentation and knowledgebases? Hah! Can you say functional illiterates or outsourced employees who are notrequired to know squat about that which they are writing about or the intended audience?
    • ZarDov2006 IBM Comment

      This is pioneer stuff that needs to be defined and understood in business terms before it evolves to a techno IT "how to" cookbook. Enterprise Architecture started within the same paradigm as SOA before it gained broad based acceptance. "Marketing" is used only when you cannot point to quantfiable results. This is one of the rare instances I have seen that a promise can be linked to an actual success story.
  • Let's do the Math

    16000 application - x = 4000 applications

    That equals 12000 applications
    There are 77 services...

    That means each service replaced approximately 150 applications...

    I'm sure SOA (in principle) is good but *that* good....

    I guess this is why this is called a BLOG and not NEWS.
    • Math is faulty

      Well, I agree with you. The math is faulty. Logically, this can not be 100% true. If a large reduction in applications occurred across the "Great Y2K Divide", I am sure that some of these sunsets happened more as a result of Y2K compliance issues (every company was faced with the same dilemna...comply or turn it off)than services. 4000 apps is still a pretty big nut to crack. I do not see 77 common services supporting a complex environment like IBM. IBM is in software, services and hardware which means Supply Chain, CRM, Fulfillment,Finance, name it...add Business Partners, Suppliers, Customers... and BAM! I did not draw the same conclusion that you did from reading the article but I do agree that it needs clarification.
    • RE: Let's do the Math

      As someone who works in the software development industry, I can testify that the number of applications that comprise a given service can indeed, if unchecked, grow to truly absurd numbers. Imagine multiple application verticals within a company each implementing its own application for registration (or geo-resolution, or email, or...), resulting in a large amount of redundant applications all comprising the same basic service. Further, over time as systems evolve and are replaced, sometimes old systems are kept around for obscure legacy reasons that never seem to get completely fixed.

      Frustrating, yes. Expensive, hell yes. Absurd, sure. But it happens... a lot...