Tuesday, October 02, 2007

herenow Complex Event Processing VI


Further continuation:

Then, the following month, I leveraged this analysis and some other CEP/ESP-related CIRs I wrote up, plus other research material that I hadn’t found its way into any Current Analysis report, into the following Network World column, which was published on August 16:



--James Kobielus

Complex event processing (CEP) has a sleek, shiny, space-age allure. CEP has been blinking on the information technology (IT) industry’s “next big thing” radar for quite a while, promising business agility through continuous correlation and visualization of multiple event-streams. Event-driven application architectures are becoming more important for modern business, as the volume of time-sensitive, real-time data that enterprise and carrier networks must process, store, and manage continues to expand.

However, CEP has yet to launch into the stratosphere of mainstream enterprise applications. For sure, the technology—also known as “event processing” or “event stream processing”--has found its niche with operational applications such as business activity monitoring (BAM), distributed process control, sensor networks, financial transaction surveillance, and integrated logistics management. But rare is the CEP application that supports the everyday needs of the average knowledge worker. CEP is still predominantly deployed as a stovepipe for specialized, albeit mission-critical, applications. And it is still primarily a vertical, industry-focused IT market segment, which is especially strong in finance, telecommunications, transportation, manufacturing, and the military.

None of which is to imply that the CEP market is not buzzing with activity or growing apace. The past several years have seen the entry of many promising, pure-play CEP vendors, including Agent Logic, Aleri, AptSoft, Coral8, Esper, GemStone, Kaskad, LeanWay, RiverGlass, SeeWhy, Syndera, StreamBase, and Vhayu. In recent months, Aleri, Coral8, SeeWhy, and StreamBase have issued important product enhancements that keep them in the forefront of industry innovation. In addition, established SOA, business process management (BPM), and middleware vendors such as TIBCO, Progress Software, BEA, and IBM have continued to beef up their CEP offerings through strategic acquisitions and product development.

But what’s conspicuously missing is any serious CEP uptake by business intelligence (BI) vendors, who could be instrumental in delivering real-time event streams to desktops, mobile devices, and other client environments. Consequently, most CEP tools must be implemented alongside users’ existing BI environments, providing a separate, event-optimized layer of visualization, dashboarding, modeling, repository, rules engine, resource connection, and administration tools.

What could explain this reluctance by BI vendors to test the CEP waters? To some degree, their wait-and-see posture reflects the slow uptake of real-time BI among their core enterprise customers. BI vendors have been beating the real-time drum for some time now. However, few BI users have been clamoring for the ability to refresh reports, dashboards, and scorecards continuously with straight-from-the-source event feeds. Many BI users can tolerate some latency in the delivery of key business data, and have been quite content to pull such data from intermediary data warehouses, which combine near-real-time data with historical information.

Nevertheless, CEP is an important complement to BI, and also to enterprise information integration (EII) solutions, which federate query/update operations directly to operational databases. It’s only a matter of time before most BI and EII vendors partner with CEP pure-plays, or acquire them outright, in order to strengthen their real-time event-driven functionality. We expect to see Business Objects, SAS, Cognos, Oracle/Hyperion, Microsoft, Information Builders, and MicroStrategy venture into the CEP arena in the next 1-2 years. Likewise, it’s very likely that a soon-to-be-independent Teradata, which has taken the lead in real-time data warehousing, will snatch up a CEP vendor to build out its real-time BI portfolio.

SOA middleware vendors will expand their CEP capabilities in order to offer event-driven architecture as an alternative or supplement to SOA. More vendors will CEP-enable their BPM environments’ BAM tools to support split-second response to changing business conditions. More enterprise service bus (ESB) vendors will invest in CEP to provide a user-friendly event aggregation, correlation, and visualization overlay to their publish-and-subscribe environments.

But the CEP market cannot achieve its full potential until the vendor community creates a consensus interoperability framework that leverages open SOA standards. One good sign is the recent founding of the Event Processing Technical Society (EPTS), a group of vendors and other interested parties that was created in late 2006 to build awareness of CEP’s applications, clarify CEP terminology, and define a CEP reference interoperability framework.

However, the EPTS has explicitly stated that it does not intend to become a standards organization, though it may work with standards groups at a later date. Unfortunately, the group has not yet produced a public draft of any such framework, nor has it attempted to reach out to groups such as the Organization for Advancement of Structured Information Standards (OASIS), which has developed a CEP-relevant standard: WS-Notifications, which supports event-driven, notification-based SOA interaction patterns.

The CEP market will evolve swiftly over the next few years as open standards emerge, as open-source alternatives appear, and as leading SOA, ESB, BI, and EII vendors acquire the most promising pure-plays. The industry’s tipping point toward ubiquity is fast approaching. By the end of this decade, the CEP arena will look very different, and enterprises will be able to deploy multi-application, vendor-agnostic, standards-based CEP infrastructures.