XML in Brief
Extensible Markup Language (XML) is a streamlined, Web-compatible version of Standard Generalised Markup Language (SGML), the complex but powerful language for defining text document formats. While its more basic Web-based cousin HTML is a fixed language that tells a browser how to display data, XML is extensible, meaning it can be used both to convey display instructions and to describe the content of virtually any file type, including Web pages, spreadsheets, database files and graphics. By allowing content to be described to a great degree of specificity, XML enables automated data exchange without requiring substantial custom programming.
When ClubComputer.com started selling computer products on the Web in November 1998, it relied on a rather quaint back-end fulfilment system. The start-up company e-mailed orders to suppliers, who manually entered the data into their own systems before processing and shipping merchandise to customers. CEO Dennis Tracz knew the system was far from cutting edge for a Web-based business, but it seemed to hold up to ClubComputer.com's daily intake of 100 orders. Then last July the company hit it big following a stint as a featured merchant on the CNet Shopper site. Volume went from $US200,000 a month to $US1 million in 10 days, burying the e-mail ordering system.
Within a month, 1000 orders disappeared en route between ClubComputer.com and its suppliers, and 12,000 customers called to complain. If the start-up was to stay in business, Tracz knew he had to find a better way.
ClubComputer.com's partners were pushing Tracz to establish electronic links using the electronic data interchange (EDI) format. But he quickly realised that EDI was out of the company's league in terms of complexity and cost. The nascent Extensible Markup Language (XML) promised the ability to establish those links using browser-based technology at a fraction of the cost and with fewer headaches. Because XML can be used to describe the content of virtually any file type - including Web pages, spreadsheets, database files and graphics - on a very granular level, the language would enable ClubComputer.com to automate data exchange without custom programming. But as with most good things, XML came with a hitch: its very ability to describe content in great detail invites different interpretations, complicating the process of data exchange that XML was meant to facilitate. The computer industry, like many business segments, had yet to agree on standards for XML vocabulary, or tags. So if even one supplier's XML rendition of a basic business term like customer differed from ClubComputer.com's, exchanging data would still require a whole lot of work.
Like many venturing into relatively uncharted e-commerce territory, Tracz couldn't risk losing momentum while waiting for these vertical XML standards to gel. Nor did his four-person operation have the resources to devote to custom programming. Tracz turned to one of the new crop of packaged XML integration tools, which handle mapping to back-end systems and pledge support for various vertical XML tag sets as they emerge. Now the company is using XML to quickly establish links to its growing supplier network. As a result, this one-year-old upstart has been able to compete with the likes of larger, more established online resellers that have deeper pockets to throw at e-commerce.
"With XML, there's no reason to sit around and wait," says Tracz, whose company is using OnDisplay's CenterStage eBizeXchange tool. "I know from experience that any sort of cross-industry agreements are fraught with politics, and I don't have time for that crap. We're a small company trying to make it in a competitive world, so we've got to move forward."
Relying on integration tools is just one of several strategies companies are using to push forward now with XML. Some of the larger shops are building their own vertical XML vocabularies yet adhering to an architecture that will allow them to modify systems to support any emerging new standards. Others are making do with what's currently available in the public domain from peers who've already delved into XML and shared their tags in hopes they'll become standards. Still others are relying on XML vocabularies published by vertical industry groups. The approaches might be different, but the message is the same: get started with XML now if you want to succeed in the light-speed Internet era.
Speak the Same Language
Why this consensus around XML? As Tracz quickly discovered, machine-to-machine data interchange is far more efficient than the e-mail, fax and phone jockeying that most companies, small and large, do when working with key suppliers or even when trying to share data internally. Although traditional EDI improves efficiency, it is typically used only by large firms and their top-tier suppliers since it requires costly and complex software on both ends. In contrast, XML delivers comparable functionality via a browser, drastically reducing the cost of entry for smaller suppliers. Furthermore, a burgeoning supply of packaged XML tools allows companies to easily take advantage of the language's capabilities without having to fight for talent in an already pressed IT market.
Yet even with what appears to be universal buy-in for XML, the lack of standard, vertical vocabularies, or tag sets, is still a major hindrance. If you think of XML as an alphabet, it's clear that without consensus on how key business terms like customer or invoice are defined, there is no guarantee that companies within the same vertical industry - let alone across industries - will treat their data in a consistent manner, hampering data exchange. For example, if one company's XML tag for purchase order is defined by a customer name and number, along with product information, but a partner's purchase order forgoes a customer number, something is bound to get lost in the translation. The problem is exacerbated when data is exchanged between companies in different industries. Taken to an extreme example, terms like sole and heel would refer to footwear components at a shoe company and body parts at a medical coverage provider.
Standard XML vocabularies for specific industries will ensure that systems exchanging data speak the same language, thereby reducing or eradicating any communication gap. "The issue of vocabulary is one of the most important questions surrounding XML today; a year ago, it was: 'What is XML?'," says Rita Knox, a vice president and research director at GartnerGroup (US). "Just because we obey the rules of XML doesn't mean we are creating messages that people outside our circles can understand."
The problem has caught the eye of industry groups across myriad verticals. Industries such as health care and finance, which depend on near real-time access to information, have already launched XML standard-setting efforts as extensions to work done by existing standards associations. Others, like high-tech manufacturing, have created new organisations to focus on this effort (see "Operating Room", page 114, and "The Same Page", page 116). In addition, the vendor community is championing a variety of horizontal XML vocabularies, many for e-commerce and other industry-spanning functions like human resources. E-commerce software vendor Commerce One, for example, promotes its Common Business Library (CBL) 2.0, an open XML spec for the cross-industry exchange of documents like purchase orders, product descriptions and shipping schedules. And competitor Ariba is pitching its Commerce XML (cXML).
Even the EDI crowd is giving its blessing to XML. In September 1999, CEFACT, the United Nations body for trade facilitation and electronic business that champions global EDI standards, and the Organisation for the Advancement of Structured Information Standards (OASIS), a non-profit consortium that promotes and oversees many of the XML vertical efforts, joined forces to foster worldwide standards for XML business specifications by sponsoring forums and publishing standards work in progress on The XML Industry Portal (www.xml.org).
In a telling show of support for XML, the automotive industry, which is so heavily invested in EDI that it may take years to evolve to an XML standard, acknowledges that existing EDI tags must be leveraged into XML. "In a sense, XML is EDI - it's just another way of transferring data between trading partners, which we've been doing," says Dave Van Noord, vice president of advanced technologies at CMI-Competitive Solutions, a Michigan-based automotive ERP provider. Van Noord also provides XML expertise and training within the Automotive Industry Action Group (AIAG). "You'll see a transition to XML in areas where we don't have a lot of EDI [tags], like for buying MRO [maintenance, repair and operations] goods. It will also enable a lot of tier-two and -three companies to be able to participate in the [electronic data exchange] process."
Most XML standards are in the early stages of development, with few industry verticals having begun any meaningful pilots. Many groups were targeting first quarter 2000 for the release of complete vertical XML tag sets. Yet because there are so many efforts under way - some competing for space in the same vertical - standard-setting progress is bound to be slow. "Our challenge is not that we get vertical industries interested in developing XML tags, it's that we manage against the multiple factions within an industry working in opposition to one another," says Laura Walker, executive director of OASIS in Boston. "It's not unexpected or unhealthy in this early stage of development. But OASIS will make every effort to unify the efforts."
In the meantime, savvy CIOs are moving forward with XML projects but keeping an eye on the standards as they develop. At Sabre Labs in Texas, that means using some of the new packaged XML integration tools where they can help and actively participating on XML standards boards like the travel industry's Open Travel Alliance, according to Bob Offutt, vice president for the R&D arm of Sabre, a global IS provider to the travel and transportation industry.
While Sabre is still functionally dependent on EDI links, Offutt says the company is experimenting with XML internally on a number of projects. XML is also a critical piece of a Sabre joint venture with IBM and Nokia, which will provide a service through travel agents to deliver Sabre travel data directly to cell phones, giving customers the convenience of checking flight information and booking hotels without having to talk to an agent or log on to the Web. That's something that wouldn't be possible with EDI, Offutt notes. "EDI is computer-to-computer, high-speed communications whereas XML's value is on the display side," he explains. "[XML allows us to] reformat Sabre data to be displayed in multiple devices."
As part of its early XML efforts, Offutt says Sabre will also define its own vertical tags with the understanding that it may have to make changes down the road. "We can't stop the car to change the tires; we have to keep on going," he says. It may be necessary to rework applications to support emerging XML vocabulary standards, but Offutt says there are plenty of tools on the market to help with that effort.
Companies should jump in by using vocabularies that adhere as closely as possible to what's been published by cutting-edge companies and industry groups, recommends Brian Lynn, vice president and derivatives architect at JP Morgan in New York City and also program manager for FpML, a messaging standard for the financial derivatives sector. And don't be afraid to contribute new work back to the evolving standard; that's a sure way of getting others in your industry to embrace what you've established, he says. So, for example, if JP Morgan has a recommendation for how a business term or process should be represented in XML, the company should pass it along to the FpML committee. Companies should also have design points in their architectures that allow for translation and conversion so that they can use XML mapping and integration tools without having to build them from scratch, says Lynn.
At ClubComputer.com, the lack of hard-and-fast vertical tag standards hasn't prevented the company from taking advantage of XML. The OnDisplay mapping tool let the company establish the necessary electronic links to suppliers like Merisel in days, not weeks. It also lets the company add electronic links to new suppliers and partners easily without having to invest substantial time and money into custom integration efforts for each new venture. And best of all, Tracz says ClubComputer.com can comfortably proceed with XML, knowing it will be able to adopt the emerging vocabulary for the computer business as it emerges. "XML is the format everyone wants to use, but [everyone's doing it] with a slight twist," he says. "With the new tools, I'm ready to integrate no matter what. I can move forward affordably with little risk."
Thanks to HL7, health-care organisations can push forward with XML Intense on-the-job training is a rite of passage for physicians; residents must dive into patient care before they've learned everything they need to know about medicine. Health-care organisations eyeing XML to build new electronic patient record systems are in a similar position. Although XML data interchange standards aren't fully developed, organisations need to get started on system design and learn as they go.
That's the attitude that Calvin Beebe is taking. The technical specialist in the information services department of a large Midwestern health-care organisation is spearheading the development of Notes 2, a rewrite in XML of an existing note-taking system that's part of the organisation's patient record system, as well as collaborating on an XML-based document management system for sharing reports. All this despite the fact that the medical industry has not yet finalised a vertical XML vocabulary. "There's a workable set of specs, and we can keep an eye open to what's going on with the standards group," says Beebe. "That lets us push forward but still remain open to change. Getting started now is only going to help."
The health care XML specs Beebe refers to are being formalised by a special interest group that's part of the Health Level 7 (HL7) committee, a standards organisation with responsibility for the HL7 messaging format already widely in use to exchange data between systems. Version 2.3 of the HL7 standard has been extended to include XML, and HL7 3.0 will support XML. In addition, the HL7 committee is working to validate a patient record architecture based on XML, which will create a standard document format for exchanging patient information. Using this standard architecture, hospitals and labs, for example, could electronically exchange test results in real time, which promises to be much more efficient than rekeying information and phoning or faxing back results days later. SNOMED International, another standards organisation in the health-care field, is also working on the effort to create vertical XML tags for different diseases and other medical nomenclature.
Without XML's ability to specify how content is classified and displayed, health-care organisations trying to exchange things like patient records or doctors' orders in HL7 have had limited success since the process requires a great deal of custom programming. And because vendors and systems integrators have had to write their own tools to handle HL7 messages, incompatibilities have hindered data exchange.
Migrating HL7 to XML addresses many of these issues and makes available a wide selection of off-the-shelf tools as well as a larger talent pool of XML experts. "XML allows us to take information from different systems, put it into a single environment and view everything that happened to a patient without going through contortions," notes John Mattison, assistant medical director of clinical information systems at Kaiser Permanente, a $US15 billion integrated health-care delivery organisation in California, and the founder and past chair of HL7's XML committee.
Kaiser Permanente plans to tap XML as a way to give its medical staff access to best practices and clinical information relevant to specific patients. With the current patient records, Mattison explains, doctors might see a reference to diabetes on a patient chart and be presented with 10 guidelines that don't reflect that individual patient's medical condition. "Docs don't want generic guidelines; they want information directly relevant to the specific clinical context of an individual patient," Mattison explains. "XML allows for a common way to represent both best practices and clinical information about a patient at a detailed level."
The Same Page
High-tech manufacturing's RosettaNet project is developing standards for industry lingo and processes If XML is the new data interchange frontier for electronic commerce, there's no industry that has caught the pioneering spirit more than high-tech manufacturing. That's why it's no surprise that this sector is one of the early trailblazers of vertical XML tag standards, the goal being to streamline supply chain operations and keep pace with the rapid-fire product cycles customers have come to expect.
The industry's RosettaNet project has made the most progress toward defining vertical XML tags of any business segment. To date, RosettaNet has marshalled widespread industry support behind XML as the standard data exchange protocol. It has also produced its own framework to define how XML documents and data are assembled and exchanged. And it has standardised on a vocabulary for defining elements that are unique to the high-tech supply chain (such as laptop or screen size) as well as more horizontal business terms like customer and part number.
In addition to defining XML tags, RosettaNet is breaking new ground by creating common XML Partner Interface Processes (PIPs), which establish standards for specific processes in the high-tech supply chain, such as how business partners would collaborate on the release and promotion of a new product. RosettaNet had released 10 PIPs as of January, and the group plans to publish hundreds by mid-year, according to Fadi Chehade, former acting CEO of RosettaNet, a non-profit organisation in California. "The high-tech sector is full of cowboys. If we're not ahead of them, we'll develop a problem that looks much worse than the tower of Babel," Chehade says. "We were lucky to start before anyone knew what XML was."
Among the dozens of RosettaNet pilots under way is one between SAP and Hewlett-Packard. To test-drive the XML capabilities of SAP software prior to production release as well as to streamline its maintenance, repair and operations (MRO) procurement process, SAP is buying HP MRO products using XML and RosettaNet standards, including a PIP for a managed purchase order, according to Christy Bergman, chief architect and development manager for RosettaNet at SAP Labs in the US. The pilot has so far yielded insights that will help SAP develop applications for collaborative forecasting and distributor reseller management. In addition to automating MRO purchasing, Bergman says the RosettaNet XML standards will expand SAP's list of potential suppliers, allowing the company to shop for the best price.
As RosettaNet continues to refine its XML tag set and adds new PIPs to define processes for everything from distribution to procurement, Chehade expects XML to transform the high-tech and electronics component industries as well as the supply chains of other verticals that have expressed interest in RosettaNet's work, including telecommunications and consumer electronics."
- B Stackpole