BeyondBItoReal.doc

Beyond BI to Real-Time Operational Intelligence
Dale Skeen

Many companies were caught flat-footed by the rapidly changing business environment of 2008, demonstrating the limitations of the investments they had made in BI and business process management (BPM) systems. With increasing globalization and linkage between disparate markets, never has there been a greater need for real-time visibility into enterprise operations and insight into the business climate as well as the ability to take action immediately. This ability to monitor and optimize daily business activities in real time is the cornerstone of true operational intelligence (OI).

Fluctuations in global supply and demand (oil, for example), customer needs (such as credit) and other variables are difficult to predict and, as demonstrated in 2008, can overtake business plans with breathtaking speed. Most companies are unable to track business and operating conditions in real time; instead, they make decisions based on data that has been uploaded into data warehouses or reports that are hours or even days old.这也是比较常见的对基于数据仓库的信息服务体系的批评，我们如何正确看待和应对？
OI和BI和EDW直接的关联关系？
Senior managers and executives spend too much time fixing, or trying to fix, problems long after they’ve occurred and the damage is done, in effect driving their business by looking in the rearview mirror.

Many midsized and large companies have invested heavily in BI technology solutions to aid decision-making and compliance. Yet those systems are inherently limited by a reliance on historical perspective and are unable to help companies effectively respond to the pace of business today. Even though many BI vendors tout real-time capabilities, BI reports may be covering data that is already 30 to 60 minutes old (sufficient time for the stock market to drop a few hundred points) or even more dated. BI systems also lack the ability to view and analyze business process data or other sources of nontraditional data.

Business activity monitoring (BAM) is another common toolset used in decision-making, yet that also fails to deliver a full analysis. BAM tools provide a real-time summary of business processes to operations managers and upper management. Originally, BAM was built on top of BPM systems and focused on the monitoring of formally modeled processes. But while BAM can perform simple analytics and key performance indicators calculations, it cannot access a wide range of data sources, perform complex analysis (such as historical, predictive and multidimensional analysis) or provide a means for someone to take action.

Moving to OI
In response to the growing need for real-time intelligence, OI has evolved to bring together existing technologies and new analytical tools to help business managers and executives keep up with the speed of business. The modern enterprise needs the ability to visualize, analyze and act in real time on operational data originating from a wide variety of operational data sources - business systems, BPM systems and nontraditional data sources - not just business transactional systems. In contrast to BI, OI relies on real-time feeds, also referred to as “event data,” which represent in real time any significant changes in data, processes, business transactions or other business indicators.

OI provides three key capabilities:
Visibility - the ability to see and access information from a wide variety of sources using a rich and interactive user interface.
Insight - the ability to analyze and draw conclusions from multiple real-time and historical data sources as the information changes.
Action - the ability to respond either manually or on an automated basis to positively impact business, processes and customers.
What makes OI achievable is the emergence of SOA, complex event processing and Web 2.0. SOA is a standards-based approach for real-time access to new types of content, especially current data from operational systems such as ERP and CRM systems. CEP is able to perform complex analytics over real-time data (known as events) in an incremental and continuous fashion. With CEP running in the background, companies cannot only see what is happening, but understand why it is happening, what it will impact and how it has been previously handled. CEP technology allows you to make correlations such as: Is a particular change in a Web promotion having an immediate impact on my best customers? Or, as my activation times continue to rise, are my customer cancellation rates rising to the same extent?

When you integrate a rich user interface with real-time analytics powered by CEP, you gain the ability to drill down into the source (system) of the information, in context. Ultimately, you gain visibility and insight into the daily operations that matter most to business success. With Web 2.0 technology, you are able to provide users with the ability to do functions such as real-time data integration that previously required IT intervention and time-consuming delays; it also allows users to blend in, or mash up traditional sources of data, such as data from transactional databases and nontraditional sources such as RSS feeds and Web-based news feeds.

What to Look for in OI Systems
A true OI system should provide access to operational data sources, continuous analytics against multiple data sources - real-time and historical - and the ability to empower users to access and act on insight in real time.
At the root of OI lies the event feeds (often referred to as “event streams”). An event feed is a sequence of changes to a data item or business condition. The simplest example is the familiar stock market ticker, which shows the fluctuating prices for stocks traded on the stock exchange. Most databases and many applications provide a notification or a trigger when a data value changes. An enterprise service bus (ESB) together with integration tools can turn these notifications into an event feed which then can be transmitted to multiple applications and filtered and correlated with real-time sources or data from data warehouses, CRM and ERP applications.

The data is made available to business and IT users via rich Web 2.0 dashboards, which can easily display information from multiple sources and visually compare real-time and historical data. Users can drill down into the context of the data and send feeds back to an event manager to trigger an automated response. Events are evaluated against relevant policies which can then automatically initiate a new business process or workflow, empowering users to take action in context.

An integrated OI system should deliver an intuitive interface for business users to define and view analytics against any of their information systems. An analyst for a shipping and logistics company, for example, should be able to ask questions at any level - from a fairly basic inquiry, such as the current on-time delivery metrics for all regions, to a more complex inquiry, such as revenue impact from missed shipments due to inclement weather conditions.

Once a query has been requested, the system should continuously evaluate the status of that query so that a user doesn’t have to keep asking the same question. Business users should not need technical training to visualize the results – instead, they should be able to click an indicator or graph on a dashboard to access deeper information and graphics. Finally, event management tools should monitor business processes in execution and provide response mechanisms when adverse events or exceptions occur, either manually or through automated workflows.

Architecture of an OI Solution
The following describes the steps in Figure 1这个图和目前业界比较多通用的SOA集成图基本一致，也是BEA的常规思路，其中ESB是企业服务总线，类似我们后续的SIP增强。
that demonstrate what happens with actionable information in an integrated OI system:

Data is accessed from a variety of sources and grouped into event feeds based on logical topics.
Event feeds are conveyed via an ESB, based on SOA standards, to other components.
The analytic server with its embedded CEP engine filters and analyzes the different event feeds, correlates with related historical data and calculates results on a continuous basis.
The results are organized into result feeds and forwarded to other components, including dashboards and an event manager.
Dashboards and reports are generated for user consumption and displayed in a standard browser.
Users can drill down into the originating system to see data details or related contextual data.
Selected event feeds and result feeds are also sent to an event manager for automated responses.
The event manager evaluates each event against relevant policies, each policy being a set of rules expressing a business objective. When a policy is matched, an associate action is triggered, which could be as simple as sending a notification or invoking a Web service, or as sophisticated as initiating a new business process or workflow in a BPM system.
During the execution of a business process, the BPM system can request additional contextual data from the original data sources.
The BPM system forwards process status as an event feed that can be analyzed, displayed and drilled into.

Life in the OI Enterprise
Traditional BI reports on outcome data, meaning the reports focus on what has happened (a sale, a delivery, a trade) for either the individual transaction or a summary of transactions. These reports generally have a relatively long-term focus (think quarterly financial statements or regional sales reports), thus the timing of these reports is not critical to the success of day-to-day operations. In contrast, OI enables “on the fly” changes to business activities because the focus is on what is happening, not on what has happened. BI typically relies on event and exception data that has been kicked out to a data warehouse or from transaction data generated by a business application, but OI also incorporates data such as RSS feeds and process data generated in real time. This enables the business to monitor and correct business processes and activities by identifying and detecting situations in need of resolution or response as they are occurring.

In a world where OI is prevalent throughout the enterprise, you can achieve service levels and improve customer satisfaction in ways that were not possible before. Now you can answer questions such as: How efficiently are business activities running? How are the unexpected event exceptions impacting business activities? And how is each business activity contributing to the business goals?

With integrated OI, retail call centers experiencing increasingly long wait times can dynamically select and transfer high-value customers to the front of the line. Airlines can automatically rebook passengers on a different flight if connections were missed. Logistics and shipping companies can dynamically reroute trucks and packages based on unexpected delays to optimize their on-time shipment rates. Energy and utility providers can match external events with upcoming weather predictions to proactively monitor demand and better manage the supply.

The core benefit of OI is collaborative and dynamic decision-making. Latency is removed, more information is shared and decisions are made when they have the most impact - while the operation is in process, to prevent problems from developing. OI enables decision-makers to respond quickly to changing business conditions in and increase operational efficiencies and overall responsiveness.