dated trip report following Gartner's 2006 BPM conference

This post is dated but still relevant. I found this while looking for some other stuff...

Following are my thoughts and perceptions from the 2006 Gartner BPM conference held in Nashville Tennessee March 26-29. The official days were March 27-29 but there were some meetings on the 26th and so I am including the 26th as a conference day.

My first observation was that Gartner’s first BPM conference was one year ago and there were around 600 people at that and within a year the same conference had grown 50%. This just highlights the growth in the BPM space.

In general, my expectations were a bit higher than they should have been. I had believed that Gartner might be publishing an updated Pure Play BPM report since their last report was published in 2004. Unfortunately, this was not to be. There are rumors that some time within the next 45-60 days their next report (or set of reports) on BPM will be published. We will have to wait and see. In all, however, I feel the conference was worthwhile and that I gained some general as well as specific insight into the BPM product space. I would have preferred more analysis on specific vendor offerings; I would have preferred more of a technical focus and less of a management or business perspective. But again, BPM does refer to Business Process Management so it is not surprising that a good deal of focus would be on business, management, and specifically on business management.

Gartner consistently says that the 1980s were the years of Quality (such as TQM, CMM, etc.); the 1990s were the years of BPR; and the 2000s the years of BPM.

A key difference between BPR and BPM is that in the BPR era, we (as in the software industry) thought in terms of “if we could just get it right”. Thus, BPR followed more of a waterfall life cycle. The focus on BPM is on agility, iteration, and adaptability.

Concepts that Gartner was particularly trying to sell include that of Programmatic Integration Server. Programmatic Integration Servers is a software infrastructure category defined by Gartner, and Gartner produces an annual Magic Quadrant for this software category. According to Gartner, "Programmatic integration servers enable a lightweight approach to providing a service-oriented architecture on top of legacy applications.”Within a single software technology, Programmatic Integration Servers enable newly developed composite applications, generally running inside application servers, to combine new business logic, with the existing business logic, screen logic, components and data present in legacy systems. Most legacy applications were not designed with integration in mind, therefore the "smarts" in programmatic integration servers center around mapping proprietary and often human interfaces to more standardized interfaces. A key consideration when selecting programmatic integration servers is the level of granularity they expose. Most legacy systems have usage profiles that define granularity from a navigation and transactional perspective which do not map well to new business logic within composite applications. The more sophisticated Programmatic Integration Servers products allow the granularity of reused legacy functions to match the needs of the composite application rather than being hampered by the original legacy application's design. The more sophisticated Programmatic Integration Servers support Service-Oriented Architectures (SOA), Event-Driven Architectures, direct data and program level access, and presentation integration support for screen-based applications. I am including two Gartner graphics which will quickly say more than I have time to explain.

One of the best sessions I attended was titled “When will the Power Vendors Offer Credible BPM Solutions?” This session was one that I really wanted other team members (i.e. Dave, Steve) to hear. Gartner evaluated the power vendors based on ten criterion: human task support; pre-built frameworks, models, flows, rules, and services; business process/policy modeling and simulation environment; human interface support and content management; collaboration anywhere support; system task and integration support; business activity monitoring; runtime simulation, optimization, and predictive analysis; business policy/rule management support (heuristics and inference); and real-time agility infrastructure supports. The power vendors named were: Fujitsu, IBM, Microsoft, Oracle, and SAP (now partnered with IDS Scheer). Gartner says that power vendors are 12-18 months away from offering a full solution but in three years could own 50% of the market. To date, the power vendor that shows the greatest capability and maturity in the BPM space is Fujitsu. Second to Fujitsu and considered a power vendor with BPM strength is Oracle. Another strong vendor that is not considered as a “power vendor” but has considerable corporate strength and viability as well as outstanding BPM capabilities is Tibco. The strongest BPM based vendors are Pega, FileNet, and Global 360. The next group of BPM based vendors with strength but less strength than the three previously named are comprised of Appian, Lombardi, Savvion, Metastorm, and Ultimus. Lastly, Gartner states that EMC, CA, and Autonomy may be interesting to watch.

Let’s digress a bit and hear Gartner describe the BPM market:

“The "BPM market" is really multiple markets or segments with over 160 software vendors vying for leadership. Most vendors are small, privately held companies with revenues under $50 million. With so many vendors, few have really enjoyed significant growth rates and few, if any, can legitimately claim leadership. As is typical of early markets, there will be multiple waves of consolidation. This will continue through 2009, as buyers become more sophisticated, and the technology matures into more-predictable sets of features and functionalities (0.7 probability)…Of the contenders in the BPMS market in 2005, we anticipate that only 25 will continue to compete beyond this time frame, with the others moving into potentially adjacent market spaces. There has already been some market consolidation through acquisition, although we expect the number of acquisitions to steadily rise through mid-2006, once IBM, Microsoft and Oracle start delivering more of their complete BPM architecture. Another indication of the early nature of this market is the tremendous diversity still in the licensing and pricing models seen and the average selling price. Through 2004, we saw many "conference room pilots" for $50,000. In 2005, we saw many more project deployments, ranging from $150,000 to $800,000. In 2006, we expect momentum to be increasing and deal sizes to get even larger…By 2007, the major infrastructure and development tool vendors (for example, IBM, Microsoft, SAP, Oracle and BEA) will deliver model-driven development frameworks and begin to challenge then-leading BPMS tool vendors for becoming the preferred platform for process modeling and design, SOA Web services development and composite solutions deployment.”

Again, I will emphasize that Gartner states that “because many BPM pure-play vendors are small companies with limited resources, no more than 25 of the 140+ competitors of 2005 will transition to the emerging BPMS market even by 2008. The rest will transition to providing more specialized tools, become suppliers of packaged processes for specific industries, geographies or horizontal processes, migrate to alternative adjacent tool markets, be acquired or cease trading (0.8 probability).

As could be expected, there was much discussion on Service Oriented Architectures. Gartner states: “SOAs and the resulting service-oriented business applications (SOBAs) have begun to give IT a method of composing new business processes from services. However, without structure, security,and integrated composition technologies, this method of composing new processes can lead to issues such as greater inflexibility, the need to maintain these new processes as custom applications, and lack of data/process integrity. In addition, this method is very IT-centric, and ultimately it inhibits super-users in building/composing their own business processes...It is also assumed that SOA enables users to buy processes from multple sources and somehow mold those into a new process. This is a false assumption, because vendors will define beginning and end points of services in different ways. Users should seek to define new processes in a controlled and manageable environment both for technology and service content.“

Gartner also states there are many reasons to use event-driven applications, rather than monolithic application architectures or even SOAs. Events can scale to higher volumes of business transactions and more users; they can simplify application development (AD) by reducing the complexity and amount of code in application programs; they can decrease the latency (response time) for a function and reduce the elapsed time for an entire end-to-end business process; they can facilitate data quality; they can enable better auditability (track and trace); they can provide earlier warning of threats and earlier notice of emerging opportunities; and, perhaps most importantly, they provide the maximum application software "plugability" to facilitate agility through continuous adjustments to business processes.

Furthermore, SOA is not as flexible or efficient as an event-driven design, but it is more flexible and agile than traditional tightly coupled or monolithic application architectures. SOA incurs the minimum possible lock-in among components in situations that involve collaboration. Companies that seek to maximize the effectiveness of their IT usage must invest in event-oriented design (as well as SOA). Their architects must have a thorough understanding of business events. They must identify and document business events in the earliest stages of business process design and follow through by implementing event-driven business components in the later stages of development and maintenance.

In the interest of time, I am going to throw out a few other observations and close with a few of the many graphics contained in Gartner’s documentation. Fujitsu made a strong presentation on standards including BPEL, BPMN, and BPDL, focusing on an explanation of the differences of purpose. In short, BPMN is about 18 months old and is simply a graphical standard whereby shapes and other graphical features used by BPM modeling tools are given standard meaning so that one can quickly glance at a model and immediately know its interpretation. BPMN does NOT imply any type of file format and it does not facilitation machine interoperability. BPEL as we all probably know is a specific machine readable programming language that facilitates machine to machine interoperability. XPDL is strong in modeling of the human elements and is used so that a model can be shared among different modeling tools. As for BPEL For People, it might be / probably will be another 3 or more years before any specification is established. Until then, it will remain a high-level concept described in a white paper. Finally, BPDM is Business Process Definition Metamodel and identifies the constrained subset of UML 2.0 that is appropriate for business process modeling. We spent the least time looking at BPDM, which is not really developed as of yet.