The hottest trend now in BPM suites is integrated performance management. While BPMS and its workflow ancestors have always featured basic process monitoring and work statistics, the kind of performance management now emphasized is more closely aligned with metrics strategic to the business: the percentage of orders filled immediately, or conformance with service level targets, broken down by customer type. These once were the sole province of business intelligence (BI) software, but performance management analytics and real-time business activity monitoring are now being brought inside the BPM suite itself. In doing so, leading BPMS vendors are trying to distinguish this new level of performance management functionality from both traditional process monitoring and the kind of deep analytics provided by BI tools.
To understand the differences, let’s first look at how these new systems work.
Business analysts start by defining business measures and relating them to process data. These measures may be based on aggregated counts, time intervals, dollar value, utilization rate, or other data types. Some aggregated business measures, called key performance indicators (KPIs), have in addition a user-defined target range or goal. Users can define notifications or automated actions triggered when measured KPI values go out of their target range. For analysis purposes, users can also define the dimensions of measurement, or how the measures can be broken out, e.g. by time period, by customer type, by product, etc. The combination of measures and dimensions defines the schema of an operational data store, a form of data warehouse that allows performance data to be rapidly recalculated, queried, and displayed in graphical dashboards.
High-level KPIs strategic to the business may involve external business data in addition to process data. The performance management component of many BPM suites is limited to business data managed or retrieved by the business process itself. On the other hand, the ability to aggregate and analyze business data and events directly from diverse enterprise systems defines the kind of deep analytics found in BI tools. A new generation of BPMS now seeks to unite these two sources of business performance data.
At runtime, the performance management component receives data in the form of events. Events are signals, typically in the form of messages, that a state change or transaction has occurred. Process events are generated automatically by the BPMS process engine upon state changes such as completion of a process activity or task. In some BPMSs, when a work item enters a task queue, is claimed or opened from the worklist, and is completed by the task participant are all considered separate state changes. The event is a signal that the state change has occurred, sometimes accompanied by context data providing additional information supporting the business measures and dimensions. In addition, external business systems can be programmed to emit events, using database triggers or integration adapters, that signal state changes such as a new or updated record in an orders table.
How those events are processed by the BPMS performance management component defines some of the differences between vendor offerings. In leading products, events are processed by a real-time event correlation engine. That engine applies event rules to update business measures with values from the event and associated context data, and applies test conditions to the aggregated results, such as a KPI drifting out of its target range. Based on these tests, the event correlation engine can trigger real-time actions – notify a manager, create a dashboard alert, launch a new process, invoke a web service, or issue its own situation event processed by other systems. This action dimension is sometimes called business activity monitoring (BAM) to distinguish it from performance management’s informational dimension, including graphical dashboards, OLAP queries, and drilldown reports, generally called analytics. Unlike BI, BPMS supports BAM with a rich platform for executing triggered actions based on performance data.
To support the analytics, the event correlation engine also populates the operational data store with event and context data in the form of OLAP cubes. These cubes – the schemas defined by the business measures and dimensions specified in the modeling tool – allow large volumes of historical performance data to be displayed in easily configured graphical charts, which break out the measures by one or more dimensions. This is the essence of the so-called “slice and dice” OLAP queries found in the performance management dashboards of most BPM suites. Generally, speaking the OLAP capabilities in BPMS are just a small subset of the deep analytics found in BI tools.
While this general performance management framework is being pursued to some degree by most BPMS vendors, significant differences in functional capabilities still exist across offerings. Areas of differentiation include:
As these capabilities become more fully developed, the strategic value proposition for BPMS will increasingly shift from efficiency and agility, where it has been centered in the past, to business performance optimization through end-to-end visibility and real-time activity monitoring. BI’s deep analytics alone can’t do this. BPMS optimizes processes with a complete round trip, from modeling to execution to proactive performance management that can take remedial action in real time. That’s something executives understand and value.
Bruce Silver
Bruce Silver (bruce@brsilver.com) is an independent industry analyst covering BPMS technology and the author of the 2006 BPMS Report series on bpminstitute.org. Additional reports and commentary can be found at www.brsilver.com.
There are no products in your shopping cart.
| 0 Items |
Membership at BPMInstitute.org includes access to our other communities
