Skip to content

More on IBM/ILOG

Lots of activity on the IBM purchase of ILOG, with most concentrating on the business rules aspect, which seems to be a driving force. A reader, in a comment, pointed to an article from InternetNews, quoting IBM’s Sandy Carter, VP of SOA (see: Ratliff was right. SOA is important!). Here is what Sandy says about optimization:

Meanwhile, IBM’s Center of Business Optimization, a group of consultants acquired when IBM bought PricewaterhouseCoopers Consulting for $3.5 billion in cash and stock in July 2002, uses Ilog’s optimization products heavily, Carter said. She noted that consultants can now more fully leverage tools such as Ilog CPLEX and CP Optimizer.

Besides not quite matching up with my memory of where IBM’s Center for (not of) Business Optimization originated (I visited Bill Pulleyblank, VP of the Center and a long time IBMer, not PwCer, and his group a couple of years ago), this description of how owning CPLEX can help IBM is a little worrying. The nice thing ILOG (and Dash) did as independents was to push optimization widely. Everyone should have some integer programming in their lives! Now, the two major optimization software systems are owned by companies that could stress consulting and services, not products.

Here are a couple of futures:

  1. The full weight of IBM’s research and marketing arms is thrown behind products like CPLEX, improving the software and selling it far and wide, or
  2. IBM realizes that it has a great competitive advantage in using CPLEX for projects that it is involved with, making it advantageous to limit the availability of the software.

I hope very much that the outcome of this is the first scenario. When I talked to Pulleyblank a couple of years ago, he very much wanted to create optimization products to sell, which is an encouraging sign for the future.

{ 2 } Comments

  1. Robert Steen | July 29, 2008 at 3:28 pm | Permalink

    You have some great points here, echoing a discussion I was having with some of my colleagues after we saw the announcement yesterday.

    As a medium-sized boutique consulting firm with an optimization practice area, we’ve occasionally thought about developing our own set of OR libraries… but we’ve mostly relied on CPLEX instead. Sure, ILOG had their own consulting group, but we rarely felt we were in direct competition with them. When we created a solution that included CPLEX, we felt like we were working with a good product and also helped ILOG sell CPLEX licenses to our clients.

    Now, with IBM as both our supplier and as competition, the game may be changing. AS a firm we will need to market ourselves better – but we should be addressing these marketing issues anyway. (How are we different from the IBM consultants? Why should a client use us when IBM has ownership of a part of our solution?) We’re always open to new ideas and techniques but now we may need to consider alternatives to CPLEX as something to position ourselves better in a changing marketplace, not just on technical merits.

    So while I’m a little disappointed (from an industry publicity perspective) that optimization is hidden deep within the press release, I’m also encouraged that IBM isn’t publically discussing this as the core of a new push to dominate “Business Optimization”. Time will tell how this all turns out! Hopefully we’ll continue to partner with the new CPLEX owners as successfully as we have in the past.

  2. Sebastian Nowozin | July 30, 2008 at 3:34 am | Permalink

    The previous poster raises a point about becoming dependent on a single product like CPLEX.

    While it might not be always possible, the use of open interfaces and standards makes it easier to switch or be ready to do so if the market changes. IBM’s has long history of support of the COIN-OR project and the Open Solver Interface, which all come under the liberal CPL license, even allowing binary redistribution for commercial purposes. OSI supports Cplex among many other solvers, commercial and open-source.

    One can hope that IBM continues to value the long-term trust resulting from an open platform like COIN-OR, which does not lead to vendor-lockin.