Enterprise Process Enablement

Toward (an increasingly) Real-Time Enterprise

Excerpted from S-Curves: Count IT in for Strategy by Max Stevens-Guille and Willem Galle. (C) 2005 IT&CA

Technology compresses time and shrinks distance. Development cycles are shorter, inventories are smaller, and teams are virtual. Customers, rapidly accustomed to change, have shorter attention spans and want their goods and services faster. To deliver, companies must invest in improving their technology and business processes to respond in real time.

Real time has been a driver of information technology in the enterprise since at least the early 1970s. SAP, for instance, was founded in 1972 with an original mission statement of empowering the real-time enterprise — the R in SAP/R3 stands for real-time. It was not until the late 1990s, however, that companies began realizing the benefits. Dell Computer’s real-time order and fulfillment system resulted in 97%+ customer satisfaction rate and helped to propel Dell to the number one slot in the personal computer industry. Until recently, non-real-time systems ran the DaimlerChysler's Mopar Parts Group distribution centers. The cost to maintain these systems and their aging “batch” designs prompted Mopar to replace them with a single real-time system. As a result, the average four-day fill rate increased from 96.5% to 98.5%, netting $20 million in savings from reduced safety stock and a $10 million savings in excess transport charges.

Real-time benefits aren’t limited to a company’s supply chain. Using a real-time infrastructure, a CFO can view the company’s financial results for the day, week, month, year, etc. Sales executives can see at a glance what is happening in their sales pipeline. Call center staff have immediate access to a caller's history supported by analytical information that can aid in cross-selling and up-selling.

With each success story, more demands are placed on IT departments to implement real-time solutions. These include everything from increased access to information, better ways to distribute information to the systems and individuals who can process it, improved techniques to gain insight from it, to better ways of collaborating — all in real-time. Sounds like old news. Yet in a recent survey of business and technical executives by Optimize magazine, only 10% of 52 respondents said their companies had a real-time software infrastructure in place.

The Enterprise Stack

With the average Fortune 500 company managing over 500 applications, it’s no surprise that plenty of opportunities exist to improve operational efficiencies. This section will introduce an overview of the enterprise IT “stack” in order to provide a context for evaluating technological trends and outline how they can contribute to operational efficiency. While simplistic, this idealized stack corresponds to current trends in service-oriented architectures.

Interfaces and Integration

At its core, the enterprise IT infrastructure is concerned with managing information. In order to achieve a real-time enterprise, if follows that one must have a real-time data infrastructure. Over the years, various elements of the database system have been in vogue. In the 1980s, databases and database servers held everyone’s attention, particularly their ability to scale to large volume batch updates and heavy query processing. The 1990s saw the introduction of complex data architectures to provide different information services to the enterprise. These included informational data warehouses, operational data stores, data marts, and multidimensional analytical databases (OLAP). Following closely behind were Extract, Transform, and Load (ETL) tools and, for a time, metadata. Next, data modeling grabbed the spotlight, only to be upstaged by business intelligence tools.

Figure 1: Simplified Enterprise "Stack.” Source: New Paradigm Learning Corporation

More recently, message brokers, B2B servers, enterprise service buses, portals and Web Services have been deployed to capture more real-time information. These technologies present new challenges for enterprise information architectures. While data warehousing tools meet many business needs, they are not agile enough to deal with real-time information. ETL tools are usually batch oriented, and the information in these data stores ranges from 24 hours old (informational stores) to up to a month (analytical databases). As an aside, this isn’t necessarily bad, for while the value of real-time data decreases over time, the value of aggregated data increases over time.

If data warehousing systems won’t handle real-time information, what will? Further complicating the real-time data issue is the variety of feeds and formats. Many feeds utilize proprietary data translation and transformation tools, metadata representations, and repositories. Once again, new technologies are necessary.

One proposed solution is Enterprise Information Integration (EII). EII creates a layer of abstraction between the applications that access the information and source systems. In doing so, EII enables a single query to access multiple back-end data sources. EII makes information available to a variety of applications in real time, and promises to simplify access.

EII is a contentious technology, which some claim will never scale. Others argue that, no matter what form the technology might take, new systems will deal with real-time information in ways traditional data warehouses could not.

Creating a standardized integration platform for real-time information can improve core business operations, reduce development costs, simplify management, and ultimately minimize risk.

Metadata

Moving up the enterprise stack, metadata enables access to relevant information across distributed, disparate information sources. Technologies that support integration, transformation, and quality of information in a time-sensitive manner are essential for dynamic enterprises. Advancements in modeling and metadata management (flexibly defining and managing information ontologies and relationships virtually across the entire data store) improve the accessibility, usability, and interpretation of information from disparate sources.

The industry is on the cusp of combining structured and unstructured data, as well as processes and other system elements, to create better metadata models. MetaMatrix, for instance, brings together three technologies to drive Enterprise Information Integration: information modeling, metadata management, and distributed query processing. Motorola is using MetaMatrix to provide information about its supply chain in real time, and Credit Suisse is using it to move away from silos toward a centralized platform.

Another vendor, Kalido, uses metadata abstractions in their tools to support dynamic data. Kalido delivers consolidated views of corporate performance without requiring systems to be standardized, and can reflect major business change rapidly.

A deeper exploration of this topic involves examining the work of the Object Management Group. More specifically, a review of their metadata standards includes: Meta-Object Facility (MOF), Common Warehouse Meta-Model (CWM), and XML Metadata Interchange (XMI). For process modeling, they use Unified Modeling Language (UML).

Companies active in this area include Ascential, EMC, IBM, Informatica, MetaMatrix, Oracle, SAS, Software AG, and Sybase.

Business Process Management

In contrast to the bottom-up methods discussed above, Business Process Management (BPM) takes a top-down approach to systems design. The goal of BPM is to capture needs at the initial design stage and make them available throughout the rest of the development process. The general sequence is as follows:

• Business professionals define needs and model business intent
• Software developers attach implementation components to the models received from business professionals
• IT deploys executable models

BPM empowers business professionals to understand their own processes. The process model and executable process are synchronized, which not only solves the perpetual problem of keeping documentation and code intact, but helps to adhere to government regulations, like Sarbanes-Oxley and ISO 900X, that require processes to be explicit and documented.

Business process management vendors specialize in either modeling or execution. As products evolve, BPM software enables a more complete process that includes collaboration and analysis.

BPM suites generally comprise data modeling, connectivity, execution run-time, and management components. At the low-end of the market, suppliers focus on workflow, automation and specific business process improvements. Higher-end enterprise BPM vendors focus on more integrated and automated systems. These products orchestrate, simplify, predict, and manage process organizations.

Pure-play BPM products are generally the most adaptable to any IT infrastructure. They automate human-oriented and paper-based processes as well as system-to-system processes. BPM systems built on Enterprise Application Integration (EAI) technology are a good bet for integrating both structured and unstructured data from critical applications and integrating that information into back-end systems. BPM add-ons to application servers or ERP systems are closely tied to their parent software and are a good fit for users of those larger products.

As companies transition their services to Service-oriented architectures BPM will become simpler. The following chart illustrates the evolution of BPM over time.

Figure 2: Migration to Service Orientation using BPM

In addition to mainstream providers like SAP, IBM, Microsoft (BizTalk), and Oracle (Collaxa), a number of pure-play BPM providers offer solutions. These include: Appian, Action Technologies, Computas, Fuego, Intalio, Lombardi, Metastorm, Savvion, Staffware, and Utimus. Computas, for instance, provides software for creating visual models of enterprise architectures and other systems to create comprehensive views of an operation for decision support and organizational governance.

For companies writing their own software, LogicLibrary provides Logidex, a tool that maps relationships among software development assets and the business processes they support, which enables customers to manage and reuse assets.

Business Process Execution Languages

As the market moves to a service orientation, new tools are required to orchestrate communication between services. The Business Process Execution Language (BPEL) specification is the cornerstone of service-oriented architecture, enabling Web Services to form business processes. Compared with traditional EAI, BPEL is a better way to integrate applications because it employs rich process semantics that make integration easier, and includes standards for extensibility.

BPEL uses Web Services to support shared tasks in distributed computing environments — even across more than one organization. Using BPEL, a programmer can describe a business process that can take place across the Web. In the supply chain, a BPEL program might describe a business process that formalizes a transaction like a product order and any exceptions that may arise.

BPEL toolsets provide adaptors to hundreds of different commercial and legacy systems. As more vendors adopt BPEL, adding and swapping compliant tools may approach the plug and play vision so heavily promoted.

Invented by IBM and Microsoft in 2002, backers of BPEL now include SAP, BEA, and Parasoft. The standard is managed by the Oasis standards body. Open source BPEL projects include activebpel.org and orbeon.com.

B-web Orchestration

Sitting atop the process-oriented elements of the business functionality stack are b-web orchestration processes — the lenses and levers through which management senses and responds to change within the organization. In addition to traditional reporting tools, b-webs use business intelligence and modeling tools to gain insight from various data sources. Figure 3 shows b-web orchestration within the rest of the enterprise stack.

Today, business intelligence (BI) systems focus on serving up information through management “dashboards” or portals that function as input devices for managers. In the long term, real-time monitoring and analytical systems will communicate back to operational systems which directly trigger changes in the business (pricing, distribution, sourcing, etc.) in response to market conditions and business performance.

Figure 3: B-web orchestration within the enterprise stack

The BI market has experienced a lot of change recently. Several factors are involved:

• Many organizations are in the process of standardizing and consolidating BI tools. As a result, they tend to choose point solutions
• The market trend is toward convergence — enterprise business intelligence systems (EBIS) and reporting delivered from a single infrastructure and vendor
• Vendors like Cognos, Microsoft, and MicroStrategy are introducing more comprehensive suites of reporting products

Enterprise Business Intelligence Systems

Until recently, robust enterprise reporting and enterprise business intelligence systems were technologically distinct. Robust reporting required complex, high-volume, and precisely formatted operational reports that were run offline in scheduled batch mode for distribution to large audiences. These tools, in effect, used a push model for reporting. In contrast, the reporting offered by EBIS was focused on simple, ad hoc, online reporting, using graphical presentations. Users interacted directly — using a pull model — with the system to perform queries and create reports.

Increasingly, driven by pressure to cut IT costs and a desire to reduce complexity and improve integration of BI tools, vendors have responded by combining the traditional batch reporting with ad hoc query and presentation into a single, interactive EBIS.

Vendors providing comprehensive BI offerings include Cognos, Business Objects, SAS, and MicroStrategy. EasyAsk Inc. is an alternative solution, foregoing the analytics to provide a search and information retrieval platform that offers single-point access to different types of data, regardless of format or data location.

Enterprise Decision Management

Increasingly, decisions must be made quickly and without the benefit of perfect information. Enterprise decision management, an elaboration of the business rules market, promises to automate at the process level. According to IDC, the concept combines:

• Feedback loops that respond quickly in predetermined ways to a changing environment
• Predictive systems that make decisions under uncertainty
• Economic models and learning that formulate, test, and implement new strategies

Business Activity Monitoring

Second from the top of the enterprise stack is the business activity monitoring (BAM) layer. With an emphasis on decreasing response time, business activity monitoring features are appearing in operationally focused business intelligence applications. These real-time functions are for frontline staff. Unlike traditional BI users, like business planners and executives who monitor the slower-moving indicators and trends, the users of operationally focused BI applications need a constant supply of fresh data. Instead of understanding the past, they must know the present. For example, a national chain of retail stores uses operationally focused BI applications to monitor the relationship between point-of-sale data and inventory.

Business activity monitoring alerts people to useful events, discovered through background processes that analyze key performance indicators for problems or opportunities.

Business activity monitoring alerts people to useful events, discovered through background processes that analyze key performance indicators for problems or opportunities. For instance, when an under-stock warning is detected, a BAM system issues an alert. Business operations’ staff can quickly research the history of the product and suppliers, and react within hours to schedule inventory shifts.

Vendors that provide comprehensive enterprise stacks, like SAP, utilize sensors to monitor the activity of various components. Sensors are modeled by their data modeler, executed in their application server run-time execution environment and front-ended by role-based alerts/dashboards. Open standards and component interoperability enable third parties to easily achieve the same function in heterogeneous environments.

David Luckham of Stanford University defines a business activity monitoring system as follows:

• provides instant insight into how IT events at any system level, e.g., network failures, database access loads, online Web site activity, and changes in metrics on all kinds of resources), will affect the progress of high-level business transactions
• permits real-time, business level decisions in response to system events, e.g., rescheduling business processes, that have stalled as a result of a credit reporting service slowdown
• automates real-time notification of violation or pending violation of business level policies or service level agreements (SLAs)
• provides statistics on business process performance

Vendors with business activity monitoring software include IBM, Black Pearl, Hyperion, Informatica, Microstrategy, Symantec, SAS, and WebMethods.

Conclusions

The most significant trend we see here is the creation of tools that provide visibility — of both underlying processes and surface issues — to enable decision makers at all levels of the enterprise to “close the loop” and reduce the time it takes to make and act upon decisions. As more of the systems that comprise the enterprise function are equipped with instruments that measure and record data, and KPIs constructed to check their parameters, management will respond more quickly to changing market and operational factors.

To achieve these goals, line-of-business management must participate in designing the architecture of their support systems. The delineation between IT and business is increasingly being blurred; traditional IT staff learns the elements of business, while business managers are trained to abstract and formalize their operational models.

The enterprise is moving toward a flexible, component-based enterprise suite that is based on open-standards, actively measured using sensors, partially automated using Business Intelligence tools, and managed by collaborative teams of line-of-business and IT staff.