The early reviews of IBM BPM 7.5 were out last week, while IBM Impact was still in full swing.? It seems that the analysts in attendance were of differing opinions about the strength of IBM's update to 7.5 - with Clay Richardson disappointed, and the other analysts ranging from reassured to impressed.
Clay's review ("IBM Adds Fresh Coat Of Paint And New Tires To BPM Offering, But Still Needs To Rev Engine") starts off:
So far, IBM is following the product integration roadmap John Rymer and I laid out in our report published immediately following IBM?s acquisition of Lombardi.
I'm sure IBM looks at it as, they were following their own roadmap and some of the points just happen to coincide with what analysts were clamoring for. One thing that the analyst community doesn't seem to be comfortable with is that IBM doesn't say much about future releases - they cite disclosure rules - and they only announce releases within the same quarter they're to be released.? But beyond that, I think it is quite right that the decision about *how* to integrate Lombardi and WPS had not been finalized at this time last year.
With today?s announcement, IBM checks off the first point of integration on our list: establishing a single repository across Lombardi Teamworks and Websphere Process Server. With Business Process Manager V7.5, IBM will deliver a single repository for process assets that leverages Lombardi?s impressive ?snapshot? version management and governance capabilities, providing a unified approach to administering and reusing process and integration assets.
I imagine that this retrofit to WPS and integration designer was actually quite a lot of work - and likely addressed the hardest technical parts of the integration of these two products.? But Clay goes on to say:
Although IBM has done a great job of delivering a unified repository, the core BPM engines and development environments will continue as standalone and separate entities ? at least for BPM V7.5. While this is not surprising ? we predicted that it would take three to four years for IBM to completely integrate Lombardi and WPS into a single unfied environment ? we expected IBM to communicate a strategy or vision for merging the engines as part of this announcement.
I think this is a distinction that won't matter to users.? It might surprise Clay to know that Lombardi, since 2005, effectively had two engines under the hood.? But it certainly never felt that way to users.? And with the integrated rules engine in IBM BPM 7.5, you could say it has 4 engines.? The point is - as long as the functionality works well together, this distinction won't matter to process authors.? There's also an option to deploy the whole stack into a single VM - particularly useful for developer machines.? Most people won't quibble over different sections of code running inside a VM.? After all, an engine is just a body of code that transforms inputs into outputs based on current state plus a model which provides context.? A good BPMS will have more than one such body of code.? Even a good rule suite will have more than one engine.
So the issue in the future isn't how many engines IBM will have embedded in its BPM suite.? The questions to ask are:
- Will future versions feel like one product or two or more products.? Clearly the direction is to make IBM BPM feel like one product.
- Will new versions of IBM BPM provide the same transformations of input to output given the same state and model context.
Information Week ran a story that reads very much like Clay's:
IBM's approach can be contrasted with that of Oracle, which took a decisive step in 2010 when it integrated the AquaLogic BPM system it acquired with BEA with its own legacy BPM product. That move yielded a single product and a clear roadmap, but it also forced existing customers of both products to do considerable migration work to move forward.
Except that when their article contrasts IBM and Oracle, it fails to mention that Oracle bought BEA in January 2008, nearly 3 years earlier (Clay, however, was more fair in his comparison).? And yet the expectation is that IBM provide this transformation in a year.
But while Clay was focused on the need to consolidate engines, others focused on the market signals IBM was sending.
As Bruce Silver wrote in his rebuttal:
Some have called it just ?a new coat of paint? on the existing offerings, because the (Lombardi) Process Designer and the (WPS) Integration Designer tools are both still there, and both runtime engines are still there as well.? But that misses the point.? Where IBM last year was pushing separate fit-for-purpose BPMSs ? something nobody really wants ? they now can offer a single BPMS that has the combined functionality of WPS and WLE.
I agree with Bruce - at a detail-level, it also ignores the interface makeover WPS Integration Designer got, to match the repository unification (which added significant versioning functionality to WPS).?? At a big picture level, it misses the point, which Bruce makes:
Beyond that, this announcement represents a major shift in IBM?s strategy for addressing the BPM marketplace.? You might even call it a palace coup:? the Lombardi/human/business-centric value system overthrowing the old WebSphere/integration/developer-centric value system, or even a BPM perspective rising above the SOA perspective.? Given the existing installed-base investment on the two sides, this is truly a wag-the-dog moment.
I think this represents IBM's move to capture the business-oriented perspective of the BPM market - something that was part product functionality, part product design, and partly go-to-market.? Bruce's summary:
And here?s the thing:? it?s ONE product.? You get it all.? Business-empowered design, what-you-see-is-what-you-execute, and instant playback.? SOA and integration services.? Powerful business rules. [...] but I think everyone is surprised they got it done already.
Bruce has another post on the BPMS Endgame which predicts that IBM will focus on BPMN2 engine work for the 8.0 release timeframe.
Neil Ward-Dutton also rebuts Forrester's assessment:
However when you look deeper, the release of Business Process Manager marks a significant departure for IBM, and warrants a thorough reappraisal of IBM?s competitive position.
He also hits on a few key points of integration:
- Unified repository toolset
- Unified governance toolset
- Single Deployment runtime foundation (no more copying EAR and WAR files around)
- Single Administration environment
Business Process Manager makes the relationship clear: Process Designer is aimed at business-facing teams collaborating to optimise business processes; Integration Designer is aimed at IT teams working to orchestrate the integration of systems to support the optimisation of those processes. Again ? these two environments work together through the use of a shared repository and governance toolset.
Tony Baer also humorously commented on the Lombardification of IBM BPM.? Unlike David Brakoniecki, I couldn't resist revisiting the analyst reviews.? David points out a few of the "unsung features" in the 7.5 release:
- A powerful REST API which in theory should allow better and richer user interfaces to be built
- A new charting technology (based on iLog jViews, I think)
I'd add to that the deployment characteristics - the fact that we will be able to build solutions with both the Process Designer and the Integration Designer - and then manage and deploy them from the same repository, to the same run-time clusters - is a big improvement over the state of the art in the previous versions.? And it appears to be a big improvement in how both WLE and WPS previously managed deployments.
Sandy Kemsley took more time to write her analysis, and it demonstrates her extra time to reflect.? I liked the shout out to our sleuthing out the announcement ahead of time (maybe IBM should include me on their analyst briefings so that we'll be embargoed as well!...).? She writes:
It?s important to look at how the IBM organization has realigned to allow for the new product release: Phil Gilbert, former president and CTO of Lombardi, now has overall responsibility for all of WebSphere BPM ? including both the former Lombardi and WebSphere BPM products ? plus ILOG rules management. Neil Ward-Dutton referred to this as the reverse takeover of IBM by Lombardi; when I had a chance for a 1:1 with Phil at Impact, I told him that we?d all bet that he would be gone from IBM after a year. He admitted that he originally thought so too, until they gave him the opportunity to do exactly what he knew needed to be done: bring together all of the IBM BPM offerings into a unified offering. This new product announcement is the beginning of that unification, but they still have a ways to go.
When the buyout happened I often heard this argument that Phil would be gone within a year.? But, living in Austin, I've seen a few promising startups purchased by IBM in my day (Tivoli and Webify just to name two), and I've also known Phil for... 10-12 years now.? My sense was that IBM has the scope and opportunity on the big stage that Phil would really relish taking advantage of.? IBM is big enough to make the right role for someone like Phil - in a way that very few companies can.? If they were willing to do it, I felt like they had a chance to hang on to Phil.? I felt the same way about most of the people acquired with Lombardi - some would leave, but IBM has the reach and size and money to keep people if it chooses (and if it acts in time).
Regarding that "two engines" argument from Clay:
However, from the customer/user standpoint, it?s wrapped into a single Process Server, so if IBM ever gets around to refactoring into a single engine, that could be made fairly transparent to their customers, but would likely have the benefit of reducing IBM?s internal engineering costs around maintaining one versus two engines.
I think Sandy hits it just right.? The issue isn't how many engines are under the hood - it is what does it feel like to the customer.? Regarding the lack of a cloud offering for BPM: "They need to rethink their strategy on this, and stop offering expensive custom hosted or private 'cloud' platforms as their only cloud alternatives."? Again, I think Sandy's right. It is hard to tell in what time frame it really starts to hurt, but the trend lines are there, and they're plain to see.
Great reviews and perspectives to soak up.? Nothing I like more than reading these competing perspectives and conclusions and then reconciling with my own opinions and the impressions of the BP3 team.