Have Data, Will Travel: A Data-Centric Approach to Enterprise Systems Development

Have Data, Will Travel: A Data-Centric Approach to Enterprise Systems Development

John A. Zumbado
GeorgiaPowerCompany
241 RalphMcGillBlvd., Bin 20020
Atlanta,GA30144-3374
Email: jazumbad@southemco.com

PhillipA. Naecker
GatekeeperSystems,Inc.
1010E. UnionStreet, Suite 101
Pasadena,CA9110-69756
Email:pan@gatekeeper.com

WilliamH. Iller
OrbitalImagingSystems, 21700AtlanticBlvd.
Dunes,VA20166
Email:mailto:iler.bill@orbital.com

Introduction-The Data Centric Approach Applied To AM/FM/GIS
The development of a large AM/FM/GIS system is always a challenge. Such systems often involve a large number of diverse data sources, with complex interrelationships between the data elements. These data sources typically include those both internal and external to the enterprise. Moreover, AM/FM/GIS systems are often burdened with very broad application requirements, including complex spatial analysis and display operations.

To make matters more difficult, large AM/FM/GIS systems often involve interaction between multiple-user departments and support organizations. Traditional application requirement definition and deployment approaches involve extended requirements gathering, business process reengineering, infrastructure upgrades, and a plethora of other activities. The result is typically a lengthy development process fraught with delayed deployment of critical information systems.

These delays often result in a failed or much under-delivered project.
This paper describes an approach that enables the rapid development and deployment of AM/FM/GIS systems capabilities in the face of such challenges. This approach has been successfully used in implementing large-scale, enterprise-wide AM/FM/GIS projects in very short timeframes.

The data-centric approach, as this new methodology has been dubbed, focuses principal attention on delivery of geospatial data and related tabular data to end users rather than providing new geospatial data or new geospatial analysis applications. This approach concentrates on identifying, collecting, coordinating, and linking together enterprise data resources, and then providing the widest possible access to this spatially enabled data via Internet tools. The business driver for this approach is that properly organized data has tremendous value to workers throughout an enterprise. Organizations gain business value by quickly providing data access instead of waiting for the development and implementation of complex applications. With the data in place and accessible, increasingly complex applications can evolve as the business need is identified and addressed.

The principal components of the data-centric approach are:

Bringing the data together
Oneofthe great stumbling blocks forlarge AM/FM/GIS applications is the management of heterogeneous and distributed geospatial and related tabular datasets that are under the control of disparate departments or external organizations. In the data-centric approach, all datasets are brought together and linked, in a single common environment convenient for "publishing" on the Internet or Intranet ("links" are both literal Web hyperlinks and less literal connections between data sources). Thus, the problem of online, realtime integration of multiple disparate datasets is reduced to a series of easily built and easily maintained data import efforts.
Staying focused on the data
Significant value within the enterprise is gained from existing geospatial and tabular data by exploiting the linkages created between the data sets. These linkages spatially-enable the tabular data sets, which facilitates spatial data integration and presentation. In order to gain this value quickly, the focus of a project must be kept on deploying the data, and the temptation to jump into developing some new application functionality.
Leveraging Internet tools and availability
Using widely available and free (or nearly free) Internet technologies, the newly organized data can be deployed as widely as possible. The entire enterprise is given access to the data, because the only tools required are Internet browsers with free or very inexpensive plugins. Because this initial access application is based on familiar browser user interfaces, training requirements are minimal or non-existant. This and subsequent applications self-install, and can be updated and extended without revisiting any desktops and generally without retraining.
Adding more applications
Even if the data is easy to access, there will be needs for applications to help the users interact with the data in more efficient and powerful ways. In the data-centric approach, such applications are developed in parallel with and secondarily to the data management task. Because the data has already been widely deployed over the Intranet/Internet in an easy-to-use environment, applications that follow can build upon a rich data environment and be more focused on specific, narrow tasks. Most importantly, application development is freed of the often complex and time-consuming tasks of providing data navigation tools, creating a deployment platform, and creating links to other datasets. Those activities have already been solved by the preceding data management and publishing tasks.

Why is the data-centric approach appropriate for am/fm/gis Applications development?
Experience to date with the data-centric approach applied to AM/FM/GIS applications development has been exceedingly positive. This success has spawned some observations as to why this has been the case. These are:

LeveraRinE Empowerment Early AM/FM/GIS applications that focus on providing complicated functionality typically require extended requirements analysis. On the other hand, a great deal of business benefit can be derived just from providing users with access to their data, and to their data integrated with other data sources as soon as possible. It has often been observed that by providing convenient and quick access to geospatial data integrated with tabular data, a significant number of end user requirements can be satisfied. This approach taps into the significant energy that personnel apply when empowered to improve business processes.
Taking the Easy Road for a Change. It is much easier to build a system that easily delivers a user's data than it is to build a system that provides new application functionality. The datacentric approach focuses on the data, providing broad and easy-to-use access to that data. Speedy Implementation. In the data-centric approach, applications can deliver access to complex geospatial data in a few months. This approach specifically rejects project schedules that exceed six months as being too risky and generally unrealistic. Typically, the business environment and software both change too much in that timeframe.
Technology Neutral. Because this approach focuses on the data and on data delivery, it is fundamentally technology neutral. Development efforts are focused on collecting, organizing, and providing access to geospatial and related tabular data, not on the application of a tool set from one vendor. Conversely, it is quite reasonable for this approach to function with multiple vendors' tools operating side-by-side.

Breaking Traditional Organizational Barriers to Data Sharing. The data-centric methodology decouples data consumers from data providers. That is, instead of a direct link between a d~~a provider and their users, data from data providers is incorporated into an enterprise data store that is then deployed to many different users. This decoupling yields a number of positive effects. Data consumers can easily access data from a variety of data producers, benefiting from the multiplicative effect of combined data sources.

Business drivers for more responsive application development
The data-centric approach has already gained acceptance in all corners of the enterprise. Analyzing organizations that rapidly adopt this approach reveals these key business drivers:

Imuroved Customer Service. Businesses everywhere are under pressure to provide quick and improved customer service. The most requested need of customer service personnel is timely access to an organizations critical business data. Unfortunately, traditional IT departments often interpret this requirement as the need for a new "AM/FM/GIS Application". The quickest way to empower employees is to give them the data now, and work over time to incrementally improve their ability to analyze and process that data.
Data Integration/Improvement. In utility industries especially, customer service and operations personnel need to integrate geospatial and tabular data to allow quicker response to customer requests and trouble spots. While it is eventually possible to build a complex data processing system that automates decision making, such systems are notoriously complex and timeconsuming to construct. Successes to date with data-centric systems implementations show that nearly equivalent levels of end-user satisfaction can be achieved. These systems are developed at a fraction of the cost and in less time then traditional approaches.
Ouick Business Operations Improvement. Traditionally, geospatial information provides only a narrow set of operations personnel, engineers and "GIS Specialists", with corporate asset data. More than ever before, businesses need to provide geospatial information to external partners, customers, and workers in order to improve and streamline business processes. This defragmentation of information, both spatial and non-spatial data, supports a total system view. With this perspective data users can see what effects their decisions have not just departmentally, but from a total systems perspective.

Mitigating risks with the data centric approach
As with other systems development methodologies, the data-centric approach has some inherent risks, but these are minimized with its proper application. Typical risks are discussed below:

Risk Number 1: Cost and Schedule Inflation
Ballooning costs and schedule creep is probably the greatest risk for any AM/FM/GIS project. The authors have found that a data-centric application development approach has a lower risk of these undesirable effects than other software development approaches, for some fimdamental reasons. First, the data-centric approach avoids complex "complete" applications. Most of the benefit of many GIS applications is achieved simply by "getting the data out" to the user. In the data-centric approach, the data is typically organized and deployed in a map-enabled Web environment Complex GIS processing systems and extensive requirement analysis are avoided.

Second, the data-centric approach is centered on deploying existing spatial data, and spatially existing tabular data. These tasks are fundamentally less risky because the data sets involved have been previously captured and are already understood. Third, a phased-in or departmental approach can be used where specific data is added or brought on-line. Departmental fimding is typically less costly and the results of the effort can be demonstrated more quickly.

Risk Number 2: Hostile Res~onse from Traditional IT or GIS Organizations
Traditional information technology organizations generally prefer a highly planned and carefully managed approach to applications development. For many, the risks of deploying an application that does not fit into the long-range IT plan are dominant decision factors. The data-centric methodology is an advocate of proper systems planning, but provides a mechanism to build upon existing company assets (data), without having to construct a complete system to do so.

Likewise, traditional GIS organizations may feel threatened by the data-centric approach. In the data-centric approach, the idea is to get the data to the end users as quickly as possible, with minimal GIS applications and tools. This can cause consternation among GIS data producers; exposing data to end-users will create additional demand for data maintenance, and potentially reducing the end-user's dependence on the GIS department for map production.

Risk Number 3: Data Source Instability
Central to the data-centric approach is the availability of quality data. One of the greatest problems the authors have encountered in deploying data-centric applications is reliability of the data sources. Other departments or organizations develop the data sources, often for completely different purposes. Oftentimes, departments that produce GIS and spatial data are not managed and staffed by data processing professionals familiar with change control and data interchange. Mitigating this risk is difficult. Experience has shown that the best strategy is to work with the other departments at the point where management structures intersect. Most departments welcome the opportunity to demonstrate to management that their work-products are being broadly used in the rest of the enterprise.

Risk Number 4: Confusion about GIS Vendors and Tools
At its core, the data-centric application development methodology is a rapid application development (RAD) technique. The idea is to quickly deploy a solution that will satisfy a large portion of the organization's demand for data. Any resistance to moving quickly is detriment to the data-centric approach. A slow data-centric development project misses the point of this approach. Software vendors, on the other hand, are interested in keeping a customer within a family of tools. If an organization wants to choose a software tool from outside their product line, vendors will encourage them to wait for a tool that "integrates fully" with existing tools and data. Vendors may attempt to insert fear, uncertainty, and doubt (FUD) into development plans.

The author's experience has shown that the data-centric approach enables data to be deployed in much less time than vendors can deploy new products. Therefore, it is recommended to adhere to a data-centric application development methodology and ignore the pleas of software vendors, and simply choose whatever tool can most quickly deploy the data. Software will come and go, but the data is the organizational asset that will be used forever.

How to do it: management, tools, candidate applications and techniques
Many people may now be convinced of the benefit of a data-centric application development methodology for geospatial information, and they may be ready to try this approach for their next (or current) AM/FM/GIS project. The authors have found that the approach works well in environments with a small number of essential elements.

Project Selection
First and foremost, the data-centric approach works best when there is a clear and very short-term requirement to address by "getting the data out". Projects where a single, bounded business need can be satisfied by providing access to a variety of data sources, both spatial and non-spatial, without the need for extensive data conversion or complex processing.

Data Modeling
Ensure that the project team is skilled and experienced in developing solid data models and rapidly implementing those models. The data-centric approach depends on a well-structured database that can link together both spatial and non-spatial elements. Therefore, one key tasks is to bring together all targeted data resources into a single, integrated environment. Remember that a "publishing" environment is being built not a data maintenance environment, so data resources should be organized around web-based deployment and integration with maps not around data maintenance tasks.

Tools
Success in the rapid deployment of geospatial data is tied to the use of extremely lightweight and widely deployable mapping tools specificallyy tools that work well within Web browsers. Previous generations of GIS tools have required too much training and were to complex to allow for simple operation by casual users. Additionally, these GIS's were too costly to be widely deployed, did not work well over the Intranet. Web-based tools that can display maps and let the user interact and query objects on the map are key to the wide deployment of geospatial data.

An obstacle to look out for is vendor-initiated "tool wars", wherein vendors argue that organizations must use one tool or another in order to maintain an integrated GIS capability within an organization. It is recommended that users utilize whatever tool will allow the organization to quickly deploy data to the entire enterprise. Experience has shown that project teams should be able to widely deploy geospatial data integrated with large relational databases in a few weeks to a month. If this goal is unattainable, the project is probably being unreasonably constrained by its soflware tools, by the skills of its development resources, or both.

Identifying Candidate Applications
As mentioned above, the authors have found that the best candidate applications for a data-centric geospatial application development are those that focus on a narrow and very short-term critical business need. Applications that are part of a grand, "strategic" vision are less likely to be sufficiently flexible to gain benefits from this approach.

It should be noted that a substantial portion of the benefit of the data-centric methodology derives from the ability to spatially enable, link and deploy existing data to a wide audience. Therefore, the data-centric approach will be much more difficult to implement if there are only limited preexisting spatial resources to support the application. In this situation, purchasing basic spatial information (such as a relatively inexpensive basic street network, often available for only a few hundred dollars) and using that to spatially enable the rest of your data is a realistic option. From these constraints, some obvious candidate applications for the data-centric geospatial application development approach may come to mind, including:

Integrating customer data within a utility organization--one can easily spatially-enable data in a utility customer information system using a commercial street network and geocoding engine. This technique locates customers, allowing users to "see" where customers are located and it facilitates "linking" customers to the utility infrastructure. Once linked, many applications can evolve, ranging from more effective outage management to proactive marketing and sales.
Enterprise-wide data access--a variety of utilities and municipal applications have been successful simply by allowing users to see facility maps, customers, parcels, and related data.

Tips and Techniques
From the authors' successes at using the data-centric approach to geospatial application development, a few suggestions are offered on how to be successfid with this approach.

Start Simple--begin with the most readily available and widely interesting data (from a business value perspective) and only the simplest of applications. Remember, most of the benefit to end users will be derived simply from being the ability to view the geospatial data in a map form and using that data to navigate to non-spatial data.
Add new data sources over time--focus on the data and not on hard-to-define geospatial data processing applications. Initial "applications" will probably be enhanced map navigation tools.
Plan to release additional data and very modest additional functionality often--it is very easy to add new capabilities to web applications, since complex application deployment is not required.
Leverage the multimedia power of the web--since a web browser can easily display scanned images, photographs, and video, use a simple map interface to navigate to and access these richer data sources.
Don't try to get the applications perfect--deliver something simple soon, rather than a much grander system at a much later date.

Case Studies
In this section, three case studies are highlighted. In each of the following systems, the authors have used a data-centric applications development methodology to successfully deploy geospatial data to hundreds or thousands of users in just a few months.

PacifiCorp's Operations Visualization System
PacifiCorp is one of the largest electric utilities in the United States, providing electricity to more than 2.3 million customers in a seven-state service area. Like many utilities, PacifiCorp has installed a sophisticated Outage Management System to capture trouble calls received by customer service agents and an automated voice response system. The system allowed operations managers to work with trouble call data to improve response time and allocation of resources.

In 1997, company management realized that the outage management system could not easily be deployed to the dozens of locations where the information was needed. The system required client software to be installed on fairly powerful computers, a high-performance network, and significant training. Company management also realized that the outage system data was the real resource. Using a data centric approach, Pacificorp was able to deploy a complete spatially enabled Operations Visualization System in four months. Pacificorp initiated the effort by purchasing a commercial y available street network and applying it to in-house AM/FM/GI S data. Service area imported from another call-management effort and customer data from the customer information system rounded out the base data sources. From the outage management system the trouble calls themselves were displayed on the maps using a very lightweight polling mechanism.

The imported data was deployed using Web technology and Autodesk MapGuide to draw maps. Web browsers minimized the installation overhead at client desktops. The new system does not provide sophisticated outage management functionality, but lets users company-wide visualize customers, facilities, service areas, and trouble calls throughout the sevenstate service area. Users can navigate to calls spatially, by call time, by service area, or by customer. They can then report on calls, customers, or a wide variety of other data. Users also create and download spreadsheets containing trouble call information, services, or any other data. The Operations Visualization System has since been enhanced to further simplify its use. Using the data-centric approach, the system is now being enhanced to import real-time facility status data from a new network control system. And the system has been enhanced with a few simple reports that provide summarized and historical data on outages.

The Contra Costa County Department of Public Works
The Contra Costa County Department of Public Works has had an ongoing project of GIS data development. The Department has also built a sophisticated and powerfil database of land ownership information, expanding greatly upon the information available from the County Assessor, and linking that expanded database to spatial information from a variety of map sources. Like most county governments, the Public Works Department has a long-term strategy for providing GIS-based applications to information workers throughout the enterprise. But that

strategy will take considerable time and resources to implement. Meanwhile, the rich store of information about land and public works facilities was locked inside the department's GIS systems. In the fall of 1998, Department management decided to use Web technology and a data-centric methodology to deploy this data to the enterprise. Specific applications to manipulate the data were not critical: they wanted instead to allow users to navigate through the spatial information, and spatially enable the tabular data sources.

The Public Works Department set up two server machines dedicated to publishing GIS and land ownership information to the Intranet. These machines were populated with data from a variety of data sources, including assessor's records, a variety of internal Public Works data sources, commercial street networks, and sources from the public domain such as census data. To date, the data is expected to be fully available to users throughout County government by early 1999.

Southern Company
The Gabrielle Energy Information System is an Intranet-based application that provides a complete system level overview of various Power Delivery organization data within Southern Company's subsidiary Alabama Power Company. The primary objective of the Gabrielle EIS is its ability to capture department "rules of thumb," and integrate these known rules into a system that provides an overall view of company operations. The Gabrielle EIS coalesces both spatial and non-spatial information to increase operational effectiveness.

Gabrielle EIS was developed as an enterprise solution, capturing data from multiple departments and sharing this data with various decision-making users. Gabrielle EIS users can research information starting with plant coal delivery schedules, down to the residential homeowner meter. The significance the data integration is in the ability to access this information so those users can determine the effects of a decision beyond the typical department boundaries. The benefits of the Gabrielle EIS is the ability to streamline and improve business processes, tighten operational coordination, reduce system administrative cost, and effectively leverage existing investments in legacy applications and databases.

Data integration and fimding evolved departmentally, making it an easier sell to upper management. Departmental allocation of $25K to $50K for a data solution developed and implemented in a few short months could be easily justified. As compared to a system wide solution costing upwards to $300K to $500K, developed over a year plus time frame.

The Gabrielle EIS was developed using common Web based tools and links, with no specific ties to any vendor development product. It was determined that current systems provide only a fractured jigsaw puzzle view whose individual pieces conflicted with each other, this contradictory data or data overlap is now leveraged by Gabrielle to improve data constancy and to accurately reflect system wide operations.

Conclusion
The mantra of this new approach to AM/FM/GIS applications development is "collect, organize, link, and publish". Complex analysis and other time-consuming and limited-value software development activities are specifically avoided. Instead, the concentration is on "getting the data" out and the first 60 to 80�/0 of the value to the enterprise. Since the data described in this paper is primarily geospatial data, Web-based map display tools and simple tabular reporting tools, linked to the objects on the map, are utilized to publish and visually interact with the data.