The Fallacy of that App Server Choice is a Virtue

Over on Developerworks, there is an article comparing Pega to IBM BPM.  I don’t know if there is much merit in dissecting the article.  It’s fairly pro-Pega but I would summarize its position like this:

If you are looking for a tool to work like Pega then you are probably going to prefer Pega to IBM BPM.

IBM BPM doesn’t and shouldn’t try to be a better Pega than Pega is.

Vendor Lock-in

One point that probably is worth talking about is the idea of vendor lock-in because of Application server selection.

For example, Pega works on multiple application servers but IBM BPM is locked down to Websphere.  Here’s the logic from the Developerwork’s article:

One major advantage i can think of for PEGA is that it can run on any J2EE compliant appserver like Tomcat, JBoss, Weblogic, Websphere, etc. The customers will have more choice and need not get locked into a single vendor.

via Developerworks

I’ve seen this question on almost every RFI in every vendor selection and hear this concern regularly.

When I was working with Lombardi Teamworks in the previous job, we ran it on Jboss and I had the same concerns when IBM bought Lombardi.  I didn’t want to use Websphere but time has changed my opinion.

Here’s why:

  1. IBM bundles Websphere with IBM BPM in the same software license and the same installer.  It’s seamless.
  2. By only supporting 1 application server, IBM has been able to simplify and streamline the configuration.  It’s not to say that this isn’t still challenging on occasion but that there is one 1 set of instructions and challenges.

If the goal is to improve ease of installation and streamline the deployment of the product, then I would like to see IBM take this approach further.  By standardising Websphere, IBM has and can continue to make a more stable product.

Among my clients only the largest who are running complex clusters require any sort of advanced Websphere skills after the initial set-up and installation.  For most use cases, I’d question that there is any significant increase in long term running costs.  Big companies already have the Websphere skills they need and smaller installations pretty much work after its installed.

In my previous life, most of the problems we had with the app server for Lombardi came from code and configuration that was specific to Jboss.  Lombardi could run everywhere but wasn’t optimised to run anywhere.

IBM has turned this problem on its head.  Websphere is an invisible part of IBM BPM and vendor selections should treat it that way.  It just doesn’t matter than much.  Vendor lock-in with BPM comes from having your process models and code locked inside Pega or IBM BPM, not from the choice of application servers.