For several years now, business activity monitoring (BAM) technology - technology that monitors key operational business events for changes and enables business managers to take corrective action - has given companies a way to gain immediate visibility into their business. As BAM has evolved, it has now become a core enabling technology for the business management initiative that gets at the heart of performance visibility: operational performance management (OPM).
OPM is the process of measuring and monitoring business activity to optimize operational decisions and improve business performance. An operational front-line approach rather than a strategic back-office approach, OPM is predicated on having real-time visibility into continually updated business information with no latency and thus relies heavily on BAM technology. This represents a significant shift from yesterday's business intelligence (BI) paradigm of using historical data for daily or weekly analysis.
OPM also differs from traditional BI in its approach to displaying information. Most traditional BI reporting and analysis systems were built to either service power users via complex analysis environments or to provide highly summarized, periodic updates to top-level management ... leaving front-line users without the indicators and controls they needed to manage business operations. In an OPM system, key business metrics are established and monitored on an ongoing basis, giving decision-makers immediate and focused insight into precisely the information they need.
In fact, an OPM system looks quite a bit different than a traditional BI system:
Figure 1: Traditonal BI and OPM System, Compared
The Need for OPM
The data warehouse and reporting tools at the foundation of a traditional BI system were designed to support strategic decision-making, not to help managers understand the current status of their business operations. As such, critical operational issues were often lost in the hour, day or week-old data fed into the system. Because while traditional BI solutions focus on what has happened in the enterprise or what might happen, OPM systems focus on what is happening at this very moment.
It's an idea whose time has come. A recent Ventana Research report asserted, "[We] expect operational use of BI to accelerate and expand in previously unrealized ways, and we anticipate that operational BI will become a major focus in global organizations in 2005 and beyond."1 Why? Line managers and operational decision-makers need instant access to critical information to monitor and manage business performance and thus are optimal candidates for implementing an OPM system. From inventory optimization to instant demand visibility to risk mitigation, OPM is a powerful promise that is rapidly gaining traction.
Figure 2 highlights how different industries might take advantage of an OPM framework.
Figure 2: How OPM Can Help Different Industries
An Architecture for OPM: Standing on the Shoulders of BAM
As stated earlier, OPM is predicated on having real-time visibility into continually updated business information, but this simple statement belies the complexity of the underlying architecture. An OPM system must include the ability to:
- Capture real-time data from a wide variety of sources - from message queues, Web services and databases to context-oriented sources such as data warehouses and traditional databases;
- Calculate temporal information by collating time-series data sets from event streams - such information can then be used to trigger time-based thresholds.
- Perform dynamic modeling by integrating event and contextual information on the fly and produce a stream of analytical models - models that can automatically update themselves based on input from subsequent event or contextual information;
- Execute business rules to set thresholds according to key performance indicators (KPIs) and other business-specific triggers;
- Provide business users with an intuitive interface for specifying business rules and viewing continuously updated metrics.
Luckily, companies implementing an OPM system can now leverage several years of BAM technology development to quickly deliver these capabilities, including such innovations as real-time data integration, temporal calculations, event correlation, rules capabilities and event-driven alerting.
The main technology breakthrough offered by BAM and at the heart of any OPM system is the streaming data store (SDS). The SDS captures business events from transactional systems via messaging, Web services or other mechanisms as those events are happening and joins this event data with information from planning systems, historical data warehouse and other operational systems in real time. The SDS materializes this combined information into composite views of the data, which provides a real-time picture of business activity. As new events enter the system, these views are automatically updated. Aggregate views of business events can also be created by layering views on top of each other, enabling users to keep running totals of specific events.
The SDS also maintains and caches "temporal data sets" for time series analysis. Fast-changing data, such as moving averages and other statistical computations, are collected into views against which time-based business models and rules can easily be applied. This capability enables continuous analysis of data that changes over time, allowing users to quickly detect trends.
With an SDS, transactions can either be pushed to the database or pulled from operational systems (in general, transactions are pushed and contextual data is pulled). As transactions occur, data from various operational and historical systems is integrated on the fly and materialized in memory. When end users view a metric or a dashboard, the information is automatically updated as new events enter the system, minimizing the risk of an end-user query impacting the operational system. In order to avoid a growing memory footprint, a streaming database expires data that is no longer required for operational analysis.
