Product placement

10 February 2006

The BPM market is crowded and confused, but navigation is possible.

Business Process Management (BPM) is a technology that embodies some fine objectives, among them the idea that it is possible to bring much greater visibility, control and order to business processes.

No one who attempts to make sense of the BPM software market, however, will find any of these adjectives apply. Fuelled by a surge in interest from investors as well as customers,

The BPM leaders

With so much confusion over positioning, identifying leading suppliers can be controversial.

In Gartner's magic quadrant for business process management, for example, there are just four companies considered worthy of a position in the top right quartile - those with both the vision and the resources to achieve long-term leadership.

These are FileNet, Staffware, Pegasystems and Metastorm. But Gartner's analysis confines itself to 'pure play' providers. By definition, that takes out IBM, SAP, Microsoft and all the EAI (enterprise application integration) vendors.

But however a BPMS (business process management system) is defined, no vendor can yet supply everything. "A successful BPMS would be a sophisticated animal," says leading BPM implementation company CSC. "The [key] capabilities are not currently provided by products available from any single software vendor."

In fact, one researcher suggests there may be as many as 300 companies offering products with a BPM label.

Although rapid consolidation is a near certainty, the market will remain confused for many years to come. Many companies will add process-flow capability to their products, while others will integrate best-of-breed products. Either way, process level integration and workflow will become ubiquitous in business software.

a category of the software market that scarcely existed in 1999 is now confused, chaotic and crowded. There are many more products, and types of products, than any diligent buyer could possibly hope to assess.

There are three good reasons for all this confusion. First, the arguments in favour of using BPM to improve corporate agility, and to leverage the value of existing investments, are very strong. That has not only encouraged a wave of investment in specialist software companies, but also persuaded many larger companies to reposition away from less attractive or less strategic sectors. "Many vendors are scrambling to buy or build BPM functionality to respond to customer demand," writes Gartner analyst Jim Sinur. Inevitably, the rhetoric of marketing often exceeds the reality of delivery.

Second, BPM, despite the purist stance of some proponents, comprises several sub-sections and functions, most of which have their own technical and commercial history. These include document management and workflow, enterprise application integration (EAI), application generator and process modelling tools, and collaboration and portal tools.

And third, while BPM can often be introduced with little disruption, the technology can, nevertheless, be complicated, involving new terms, difficult concepts, and new corporate software architectures. Although most analyst companies, and most leading suppliers, have put forward clear definitions of a BPMS (business process management system), these do not map easily onto each other.

The result: "At a time when clarity is more important than ever, BPM-related terms have become increasingly confusing," says Gartner.

What is a BPM system?

There are several competing descriptions of what makes for a BPM system. The objective is to provide a management platform for storing, managing and manipulating automated business processes, whether they are long or short, or whether the underlying functions are carried out by systems alone, by people, or by a combination of the two. And it should be possible to change the processes at a graphical level without having to recode.

Jon Pyke, Staffware's chief technical strategist, lists five functional elements that a handful of leading products are able to perform today. These include:

A process definition or modelling tool for mapping out and defining a process. Some high level modelling tools may be too high-level to help with building an 'application'. This tool may generate a proprietary script or a BPEL or BPML-conformant set of instructions for the process engine to execute.

An execution engine. This is a critical component - it is the core database and engine that holds all the rules set out in the process definition tool. This automatically initiates and manages processes carried out by applications, and also manages work that is routed for human intervention. Critically, the run time engine maintains the status of all the processes.

An analytics, or business activity, monitoring tool. This is to enable managers to improve process handling, to see where there are bottlenecks and to monitor work in process.

An integration layer. This involves linking into all the business applications - not just packages such as Siebel or SAP, but also legacy systems, such as IBM CICS applications. It may also involve embedding a process engine into an application - the reverse of the usual process. This integration layer may also handle the 'orchestration' of web services.

A client interface for people to interact with the process engine, in much the same way that an application might.

There are other ways of assessing the key features. Gartner, for example, suggests that 'in-flight' agility features - the ability to change processes on the fly, as well as analytics - are important. The Business Process Management Initiative (BPMI), a BPM standards body, tends to focus more on the ambitious end-goal of BPM - the ability to handle and deploy automated processes just as if they were data components.

How well do available systems meet the requirements? This will always be debatable. Andy Bailey, marketing director of Metastorm, argues that pure-play BPM companies

BPM's heritage

BPM has a complicated history, with many antecedents, as the adjacent chart from Forrester Research shows.

All of these antecedent or component categories have one thing in common: they all address the problem of automating processes beyond and between application or functional boundaries.

EAI, for example, is concerned with linking applications through data connectors and process adaptors, so that a superset application can be created. This superset may be an end-to-end process, such as ordering a part, checking its availability and details, and arranging for delivery and settlement.

Workflow is another core element. This is concerned with automating and managing a process, such as handling an insurance claim, and has been especially effective where it involves human intervention or exceptions that can't be easily automated. Modelling tools, which have been used by software analysts for years, also play a key role.

BPM, however, is more than all this. It brings all these things together and brings in manageability, scalability and reliability. "It's the core state management tool. IT takes a graphical description of a process and executes it. It's the traffic cop, directing activities, says Chris Phillips, chief marketing officer of Staffware.

are more likely to perform all these functions, since they do not come from, for example, an integration or a workflow background.

An example: EAI vendors, such as Tibco and WebMethods, might be very good at integrating automated applications into a composite process workflow, but are less capable of managing processes that are long, or contain many exceptions, or that involve human intervention. "There are fundamental problems in taking an event-driven message bus and integrating human events," says Shawn Price, CEO of Savvion, a pure-play BPM vendor.

EAI vendors do, of course, contest this, just as those from various other backgrounds will claim that they can perform BPM functions well. Vitria, WebMethods and Tibco all have plans to radically improve their support for human processes over the coming years.

There are two other categories of vendors that are likely to press their claims for inclusion on many BPM supplier shortlists. These are the platform vendors - Microsoft, IBM, Oracle and BEA - and the big application providers - SAP, PeopleSoft and Oracle. "The platform companies have some good advantages. I view them as potentially very significant in the near future", says Chris Phillips, chief marketing officer of Staffware. However, of this group, only IBM has a set of products meeting the above criteria, and analysts portray these as fragmented and more development oriented.

The application vendors, with their new-found ability to build cross-functional workflows (using tools such as, in the case of SAP, Netweaver) are also winning some business, and are certainly making loud marketing claims. But many customers will be sceptical, says Phillips. "Why go to an application vendor when the BPM layer is separate?" he says.

CSC, which supplies its own prepackaged deployment architecture using best-of-breed solutions, has some critical advice for customers looking at all these competing claims: "Do a proof of concept. If the solution does not work in a few days, it is not likely to work within the additional days."

A new market

One of the tactics that suppliers have adopted in order to increase the appeal of their software is to package up some of the more commonly used processes that their customers use and re-sell them as 'process templates' or 'process accelerators'. This is an intriguing, and possibly highly significant development, since it points to the birth of a new market in 're-usable' business processes. However, it is still early days, and once again, the market and technology is confused.

There are, arguably, three different types of re-usable processes that customers might purchase.

The first type is perhaps better described as a 'composite application', although suppliers are using the term 'process accelerator'. These tend to be built using one of the big enterprise software packages, and involve making tight process-based links between various component applications. Several tools and standards are being introduced to support these - including Commerce One's Conductor, Siebel's UAN (Universal Application Network) and SAP's Netweaver - for linking up component functions to form composite business processes. EAI vendors are also active in this market, offering dozens, or even hundreds, of pre-packaged processes.

The second category of re-usable business processes contains components that are more like BPM templates - commonly used business processes that have been pre-configured by suppliers for customers in certain key sectors. Staffware, for example, has 'process frameworks' for insurance claims processing, mortgage processing, telecommunications provisioning and other popular areas; Savvion has 'strawmen', which perform a similar function; and most suppliers have equivalent solutions.

The third category contains re-usable processes that can be accessed and traded, possibly by different BPM systems, over the Internet. These are much talked about, but don't really exist beyond the development labs. Ultimately, these processes will slot in alongside re-usable objects and web services as IT's currency of the future.

Information Age: News, analysis & insight for IT & business leaders