From EDI to XML

For the past 25 years, Electronic Data Interchange has been the only real method for electronic transmission of business data between buyers and sellers. As we have already seen, however, although many large companies moved to develop EDI connections at least between their major trading partners, the combination of high maintenance overhead, expensive leased lines, and cumbersome protocol translations meant that EDI was too complicated and too expensive for all but the largest buyers and their key suppliers. It has always been prohibitively expensive for small firms—suppliers or buying companies. It is safe to say that any company that can afford to have an EDI link probably already has it. In fact, only 300,000 firms worldwide have adopted EDI, and even retail industries, where fast-moving goods most merit this sort of electronic exchange of business information, have fewer than 20% of their suppliers connected.

In 1996, however, a potentially revolutionary business data interchange standard came to the market with a new Extensible Markup Language (XML), which promised for the first time to provide a simple and affordable solution for secure exchange of transactional business data between firms.

XML was everything that EDI was not. Because EDI assumed (correctly) that bandwidth would be extortionately expensive, EDI messages use a compressed and confusing set of codes, and much of the explanatory metadata that help programmers to decipher and debug messages has been left out. These compressed message formats not only make EDI transactions expensive and difficult to code, but they are inherent to the code itself, which essentially means it can’t be fixed. All of this makes EDI programming difficult and expensive, and EDI programmers often difficult to train and to keep. This is particularly true today, when EDI programmers—with a wary eye on XML—are becoming increasingly concerned about being trapped in an old technology that may become entirely obsolete in a matter of months.

XML is not a language itself, but a meta-language standard that provides a flexible and inexpensive way to create common data formats. A subset of the Standard Generalized Markup Language (SGML), XML uses plain text to provide tags that describe both the format for the data and the data content itself. These tags can be used to easily identify key pieces of everyday business data—an address, a price, or a customer name—and to code for a transfer of that data to respective symbols in other applications. This means that once XML tags are programmed to recognize and match against another application’s symbols, that application can continuously receive transfers of data without having to redefine these links. Therefore, if all suppliers use an agreed-upon XML standard, once the interactive format is defined, the in-house system can read electronic data messages from any supplier using that set of XML data tags.

To make things even more sensible, each XML document is self-defining and carries with it a Document Type Definition (DTD) that provides an explanation of the data language used in the document. In this way, although XML does not affect the way that companies label or organize their current data, it means that any system that supports XML can read and understand the data inside the document.

Creating DTDs is not difficult, and unlike Java and other codes, XML is easily learned and manipulated. But although each company could reformat their current systems to understand various DTDs, a proliferation of various and overlapping DTDs helps no one. For that reason, many hundreds of interested companies, including key IT industry leaders such as Sun Microsystems, IBM, and Microsoft are moving quickly to set industry standards for DTDs that are universally accepted by suppliers and vendors in their industry verticals. For those companies that have significant investments in EDI, there are a number of translators that will, with varying degrees of effectiveness, attempt to transform EDI into an XML format, or to break out the EDI codes into readable XML symbols, using an EDI parser.

In 1998, the Data Interchange Standards Association (DISA) conceded that XML as a Web-based technology would very likely replace the traditional ANSI X12 EDI as the business-to-business standard for business data exchange. For buyers, this means that they suddenly have potential direct electronic access to secure business file data transfer from small or specialist suppliers who would never have been able to participate in a program of EDI. For suppliers, they now have a relatively simple and inexpensive way to communicate directly with buyers.

However, as always, the devil is in the details, and there is still no true agreement on cross-industry standards for protocols at the product labeling or business transaction levels, and many groups within each industry are still struggling to come to terms with different variations. For example, if your company is in the financial services arena, it is likely that your agreed XML standards will be based on those agreed to by the 6,800 member banks of the SWIFT cooperative (Society for Worldwide Interbank Financial Telecommunications). Electronic component suppliers will use XML data formats agreed to by RosettaNet. Insurance firms will use ACORD standard protocols. If as a supplier, you sell horizontal product lines—office supplies or travel services—to companies in a variety of industries, you may need to adhere to standards set by members of the OAGI (Open Applications Group, Inc.). Exchanges built on Ariba’s software use the cXML (Commerce XML) protocol, while those using Commerce One are based on xCBL (XML Common Business Library).

Accordingly, the greatest obstacle remaining is that (as of 2001) there are multiple proposed XML protocol standards under development through major industry collaborations, and although they all profess to have the same goal—to develop well-accepted schema, or vocabularies, that will provide standardized and predictable dictionaries of XML terms, and repositories that will store and help manage the product descriptions—they are not all necessarily working in harmony. These include:

• CommerceNet, a business consortium, which is developing eCO Framework X.

• RosettaNet, a consortium of electronics companies with a supply chain focus, is possibly the most advanced and influential of the several standards groups, and membership includes powerful enterprises such as Federal Express, Cisco, American Express, and EDS. RosettaNet has not only developed XML standards for product catalogs, including a dictionary of 3,600 terms describing things such as components, parts, and finished IT products, but has also devised over 100 XML-based business processes. Its goal is to produce a full supply chain XML standard for the IT industry.

• Vendors such as Ariba and Commerce One also are defining their own XML schema, or industry vocabularies, as they build their online trading communities. Commerce One has created a common business library (CBL) as part of a government grant from the U.S. National Institute for Standards and Technology and has made its DTD repository available through its Web site. Ariba, too, has put a great amount of energy into supporting commerce XML (cXML) and into developing its XML Interoperability Bus Architecture, a framework that will support third-party software integration based on standard XML interfaces.

• The Open Buying Initiative (OBI) is a standard for Internet-based business-to-business purchasing that is sponsored by a group of large buying organizations and their suppliers. Mostly focused on indirect materials, the idea was to provide a mechanism for incorporating EDI within OBI objects, but is in reality fairly limited, and few suppliers have moved to adopt the standard.

• The BizTalk framework, sponsored by Microsoft, helps to convert business objects from other applications into XML and is closely linked to another Microsoft creation, the BizTalk.org Web site that also acts as a repository for XML schema. Members include Boeing, the Open Applications Group, some of the most influential ERP vendors—Baan, PeopleSoft, and SAP—as well as Ariba and Commerce One.

• The Organization for the Advancement of Structured Information Standards, a nonprofit and vendor-neutral consortium, also has plans for its own repository of XML schema, accessible on a Web site called XML.cor.

• The Open Applications Group, Inc.—members include Lucent Technologies, Ford, Microsoft, and IBM—is a nonprofit consortium focusing on best practices and process-based XML content for e-business and application integration. It is the largest publisher of XML-based messages for business software interoperability.

Although there are some who lament the end of EDI and still feel more comfortable with the security of the direct and proprietary link between buyers and sellers, few can doubt that XML is the way forward. The greater hesitation, for many companies, is to abandon what for large companies can be multimillion dollar investments in these EDI links. Nonetheless, the writing is on the wall.