• Systems Engineer (Senior)
• Project Manager - IT Relocation...
• Desktop Team Leader
• NATIONAL SALES MANAGER. WIRELESS...
• National IT Manager

BAM to speed app reports

Carolyn April, Information Age

10/12/2002 17:14:00

BAM to speed app reports

As business process integration strategies march up the architectural stack, technologies are emerging to address growing concerns about how to manage application transactions running across distributed systems.

Vendors, including Microsoft, Tibco Sofware, Sybase and webMethods, have joined others in the BI and data-warehousing arenas, to become proponents of a concept that analyst firm Gartner has dubbed BAM (business activity monitoring). The moves represent what analysts see as a growing enterprise need for improved business process orchestration technologies.

BAM, at its simplest, represents a technology layer that sits on top of a BPM solution to capture and analyze business process data in real time.

Events are presented to the user via a dashboard that offers instant insight into such business metrics as CRM call centre success rates or supply-chain inventory levels. Evaluation cuts across multiple applications, data formats, and systems, analyzing boundaries between apps where business processes intersect.

Real-time alerting to breakdowns in processes such as shipping schedules lets users react to problems on a dime.

Application integration vendors are well-positioned for BAM because their integration platforms touch all parts of the enterprise, making the ideal data collector.

Microsoft's efforts include plans to introduce BAM capabilities to its BizTalk, Content and Commerce servers, which are due for integration under the Jupiter moniker in 2003.

Jupiter's BAM capabilities will give business users the ability to monitor businesses in real time, said David Wascha, Microsoft's BizTalk Server product manager.

Tibco recently scooped up pure-play BAM company Praja. Praja's technology sports the capability to peer back into historical enterprise data to identify patterns and information gaps, and draw comparisons to events happening in real time, which is another hallmark characteristic of BAM, according to Gartner.

In July, webMethods and analytics titan Informatica ratcheted up an existing BAM partnership, promoting their respective products as a unified solution for capturing and analysing data on business processes in real time.

And for its part, Sybase is readying a new version of its integration broker, code-named Ohio, which, among other things, beefs up BAM capabilities in its BizTracker product for real-time business activity monitoring, alerts, and notifications. Ohio is slated for release during the first quarter of 2003, according to Mike Gordon, senior manager of integration product management at Sybase.

"BizTracker lets you roll up message-level information from three to four systems and measure them against higher-level metrics, such as how many new purchase orders came in today," Gordon said.

Other vendors, including SeeBeyond and Vitria, have also made BAM-like monitoring part of their integration suites.

BAM will become necessary for enterprises looking to exploit investments in expensive integration middleware and BPM software, according to Gartner analyst David McCoy.

Gartner predicts that the BAM market will not hit full stride until 2004 or later but suggests that companies lay the groundwork now. "You have spent a lot of money on your integration infrastructure. BAM lets you justify those purchases to the business folks in your company by giving them a way to keep track of performance," McCoy said.

Many companies think BAM looks great on paper, but they're not convinced it's ready for prime time.

"This is pretty conceptual right now, and I think the time frame of 2004 for this happening is pretty unrealistic," said Saurabh Sachdeva, an SAP integration specialist for systems consulting company Xansa. "I think this is the next level of [process integration], but it only works now for large companies that have a very stable integration environment."

Meanwhile, other analysts such as Darcy Fowkes, research director at Aberdeen Group, believe that real-time process monitoring will become a stock-and-trade component of BPM software, rather than constitute its own market.

Real-time business intelligence

What technology advances will be required to make BAM (business activity monitoring) a reality? In truth, the vision of a real-time intelligence layer may be more dependent on changes in human behaviour than on technology advances. For BAM to work, for example, people must agree upon and then input the metrics by which they want to run their business. That's a tall order for most enterprises.

But assuming managers are willing to input these metrics and tolerate a "push" model of management information (vs the traditional "pull" model of business intelligence reporting), several key technology enablers will have to fall into place before BAM can truly achieve its cross-enterprise promise, according to executives we spoke to at IBM, webMethods, and elsewhere.

First, EAI vendors must tackle the integration scalability issue to deliver the ability to collect and work with terabytes of data in near real time. They must enable enterprises to move much larger volumes of data through their architectures, and figure out the right way to architect systems that take best advantage of both traditional batch-mode ETL (extraction, transformation, and loading) and newer, real-time processing techniques.

Second, database vendors must optimise their platforms to support ad hoc queries and analyses from potentially thousands of frontline and executive-level desktops. Expect to see continued advances from these vendors, for example, striking a balance between traditional modelling and load-optimisation techniques and performance optimisation on the data access side.

Third, data integration vendors will have to provide better tools to deal with data quality issues to combat the age-old problem of garbage in, garbage out.

Fourth, analytics vendors and customers alike will have to tackle the metadata management problem, with better data-standardisation concepts and better tools for storing data definitions and making them available throughout the enterprise.

And fifth, software frameworks will have to be developed to tie together the multitude of data marts and reporting systems that have proliferated throughout most large enterprises in recent years.

But don't expect a vendor rush to solve these problems all at once. A lot of the pieces and parts are there, but it's not clear the demand is great enough to justify large investment in connecting them end to end.

- David Margulius