CASE
Apower consortium that distributes a mix of "green" and conventional electricity is implementing an XML-based settlements system that drives costs out of power distribution. The Northern California Power Agency (NCPA) is one of several state-chartered coordinators in California that schedules that delivery of power to the California power grid and then settles the payment due to suppliers. NCPA sells the power generated by the cities of Palo Alto and Santa Clara, as well as hydro- and geothermal sources farther north. Power settlements are a highly regulated and complicated process. Each settlement statement contains how much power a particular supplier delivered and how much was used by commercial vs. residential customers, and the two have different rates of payment. The settlements are complicated by the fact that electricity meters are read only once every 90 days; many settlements must be based on an estimate of consumption that gets revised as meter readings come in.
On behalf of a supplier, NCPA can protest that fees for transmission usage weren't calculated correctly, and the dispute requires a review of all relevant data. Getting one or more of these factors wrong is commonplace. "Power settlements are never completely settled," says Bob Caracristi, manager of power settlements for NCPA. "Negotiations over details may still be going on a year or two after the power has been delivered." Furthermore, "the enormity of the data" has in the past required a specialist vendor that creates software to analyze the massive settlement statements produced by the grid's manager, the California Independent System Operator. NCPA sought these vendor bids three years ago and received quotes that were "several hundred thousand dollars a year in licensing fees and ongoing maintenance," remarks Caracristi.
The need for services from these customized systems adds to the cost of power consumption for every California consumer. Faced with such a large annual expense, NCPA sought instead to develop the in-house expertise to deal with the statements. Senior programmer analyst Carlo Tiu and his team at NCPA used Oracle's XML-handling capabilities to develop a schema to handle the data and a configuration file that contained the rules for determining supplier payment from the data. That file can be regularly updated, without needing to modify the XML data themselves. In doing so, the NCPA gained a step on the rest of the industry, as the California Independent System Operator started requiring all of its vendors to provide power distribution and billing data as XML files. NCPA has already tested its ability to process XML settlement statements automatically and has scaled out its Oracle system to 10 times its needs "without seeing any bottlenecks," says Tiu. Being able to process the Independent System Operator statements automatically will represent huge cost savings to NCPA, according to IS manager Tom Breckon. "When settlement statements come in," Breckon says, "NCPA has eight working days to determine where mistakes may have been made. If we fail to get back to [the California Independent System Operator], we lose our chance to reclaim the monies from corrections." Yet, he acknowledges, "we can't inspect that volume of data on a manual basis." Gaining the expertise to deal with settlements as XML data over the past three years has cost NCPA the equivalent of one year's expense of a manager's salary. Meanwhile, NCPA has positioned itself to become its own statement processor and analyzer, submit disputes to the California Independent System Operator for corrections, and collect more of those corrected payments for members on a timely basis.
"In my opinion," says Breckon, "everybody will be doing it this way five years from now. It would reduce costs for all rate payers." In the state of Ohio, almost 1,000 police departments have found critical new crime-fighting tools by gaining access to the digital records kept by neighboring law enforcement agencies. The Ohio Law Enforcement Gateway Search Engine is an Internet-based tool that can securely comb through numerous crime databases using a single log-in and query, making it easier to use than separate crime databases. For police officers, searching for information on a suspect or a rash of crimes used to require manually logging into several separate crime databases, which could take hours. Now, officers in even the smallest communities can log in just once and quickly gain access to criminal information. The project, which began in 2003, faced a major hurdle: finding a way to get the disparate crime information systems to interoperate with each other. "Everybody wants to share, but nobody wants to use the same product," says Chief Gary Vest of the Powell, Ohio, Police Department, near Columbus.
In a major metropolitan area in Ohio, there can be 30 different police departments, each using different products that aren't linked, he says. "That made it difficult for local departments to link suspects and crimes in neighboring jurisdictions." To make the systems compatible, crime records management vendors rewrote their software so that data from participating departments could be converted into the gateway format for easier data sharing. The vendors used a special object-oriented Global Justice XML Data Model and interoperability standards developed by the U.S. Department of Justice for such purposes. What makes this project different from other fledgling police interoperability programs in the United States is that it's a standards-based system.
"You don't have to throw out your vendor to play," notes Vest. So far, Ohio police can't search on criminal "M.O.'s," but that capability is being worked on. By combing local police records, officers can search for a suspect's name even before it's in the national databases or other larger data repositories, says Vest. "You're a step earlier." Other regional police interoperability projects are in progress around the nation, but this is believed to be the first statewide effort. In San Diego County, police agencies have been sharing crime data for 25 years using a custom program called the Automated Regional Justice Information System (ARJIS). Barbara Montgomery, project manager for ARJIS, says it differs from the Ohio initiative because it is main frame based and all police agencies have to use the same software to access information.
Such data-sharing programs are not widespread in the United States because of their cost, especially for smaller police departments, she says. In fact, ARJIS was made possible only after a number of departments pooled their money. "No single police department could afford to buy [the hardware and the skills of] a bunch of computer programmers so it was truly a ‘united we stand, divided we fall' approach," Montgomery says. "The next generation of ARJIS is being planned now, with the system likely to evolve over the next few years from its mainframe roots to a server-based enterprise architecture for more flexibility," says Montgomery. Along the same lines, the Florida Department of Law Enforcement will begin work on a $15 million project to inte grate the back-end systems of 500 law enforcement organizations across the state.
In many cases, investigators in Florida law enforcement offices now gather information from other departments in the state via telephone or e-mail. The Florida Law Enforcement Exchange project promises to provide access to statewide law enforcement data with a single query, says state's CIO Brenda Owens, whose IT unit is overseeing the project. "Our goal is to provide seamless access to data across the state," says Owens. "An operator sitting at a PC in a police department doesn't know or care what the data look like; they can put the inquiry in and get the information back." Large integration projects such as this often derail because it's difficult to get different groups to agree on metadata types. "The metadata management or understanding the common elements is a huge part of [an integration project]," notes Ken Vollmer, an analyst at Forrester Research. "Trying to combine information from two agencies-that is hard enough. In Florida, you're talking 500 agencies, and they have to have some software to help them determine what the common data elements are."
CASE STUDY QUESTIONS
1. What is the business value of XML to the organizations described in the case? How are they able to achieve such large returns on investment?
2. What are other ways in which XML could be used by organizations to create value and share data? Look for examples involving for-profit organizations to gain a more complete perspective on the issue.
3. What seem to be important elements in the success of projects relying on extensive use of XML across organizations, and why? Research the concept of metadata to inform your answer.