In a panel discussion at Software Development 2004 entitled "Marrying SQL, XML, Web Services and Grid Computing", Peter Coffee of eWEEK magazine best summarized the state of these four technologies when he said that customers would not abandon SQL, wanted to use XML to repurpose data, and considered Web Services the default implementation strategy for many projects. On the other hand, they felt grid computing was still three to five years away.
The panel consisted of Jim Melton (Oracle), Rick Cattell (Sun), Daniela Florescu (BEA), Jim Gray (Microsoft), Jim Kleewein (IBM), and the aforementioned Peter Coffee. It was moderated by Ken North.
The discussion ranged over a variety of topics, from the impact of Open Source software to the relevance of standards bodies, but the biggest questions were the tension between XML and SQL, and whether Web Services were here to stay.
With respect to SQL and XML, the panel agreed that SQL was here to stay. SQL, said Coffee, represented a "massive foundation" of software, data, and skills, and that customers had no reason to abandon it. Put more bluntly, Jim Kleewein said that there was a multi-trillion dollar investment in SQL and that "anybody who says that SQL is dead is [themselves] dead" and was "out of touch with reality."
The question became more interesting when viewed through the lens of data integration -- specifically, whether integration would continue to be done with SQL or would instead be replaced by XML.
Kleewein saw this in SQL-centric terms. There would, he contended, "continue to be a large market for SQL-related integration functions" that would "continue for the foreseeable future." This would be augmented by XML capabilities, which would allow users to represent, in SQL, data that can't be represented using SQL's primitive data types.
Dana Florescu disagreed, saying that SQL could not be used in any of BEA's data integration use cases. The problem was that all of these cases included at least one data source that could not be mapped to SQL, such as native XML databases, Web Service, LDAP, or XML itself.
While Kleewein dismissed this as "the object-oriented cry of today", Florescu countered that this was "very different", saying that XML (and XQuery in particular) was much more natural for use in integration. Furthermore, it was much more likely to be implemented correctly than SQL. Kleewein countered this by saying that XQuery implementations had not been around long enough to find the discontinuities between them.
On Web Services, there was much more agreement. Ken North opened the panel with a rather impressive set of statistics. IDC forecast the Web Services market to be 34 billion dollars in 2006 and noted that the average Web Services project was relatively small, costing just 50 thousand dollars and lasting only 90 days. Another survey found that 85% of enterprises were planning Web Services projects. Finally, InfoWorld reported last August that 51% of surveyed programmers were currently working on Web Services.
In response to a question from the audience as to whether Web Services were a niche solution, Peter Coffee said they were not. While there were some concerns about Web Services, such as security, transactions, and incompatible data models across enterprises, they had become the default processing model for in-house projects, a point with which Jim Kleewein concurred. Coffee also noted that Web Services were common in situations where there was an "asymmetry in bargaining power," such as between Wal-Mart and their supply chain.
In the lone disagreement, Dana Florescu felt that it was not easy to build Web Services, that they cost a lot of money, and that the tools weren't necessarily there yet. Jim Gray countered this by saying that, while he had been unable to make DCOM work and had had projects fail as a result, he could "make Web Services in a heartbeat." He also agreed that Web Services are common inside the organization and that they were used by "almost everybody who has an outward-facing model," such as Kinko's, eBay, FedEx, UPS, and Amazon. Furthermore, while many of these services had been experimental, many of these experiments had "taken off."
Two lesser topics of discussion were Open Source and the future of standards organizations. Dana Florescu felt that Open Source software had had a "huge impact," noting that almost all W3C recommendations had implementations in Apache and that these implementations were of good quality. This, she felt, was one of the reasons that W3C recommendations had fewer interoperability problems than the SQL specification.
Peter Coffee concurred, noting that the credibility of Open Source products was quite high. A year ago, his readers had drawn a line between Open Source and commercial products, but now they treated all products as being in a single pool and made choices based on quality, support cost, and so on.
While Jim Gray felt the Open Source community had been very helpful, he was puzzled by the business model. How would there be a software industry if there was Open Source? He answered his own question by stating that people selling software had to be better.
Jim Melton agreed, noting that Open Source products forced commercial vendors to write better software. In answer to a separate question, he felt that Open Source databases did not threaten commercial databases, but did force them to compete on terms other than price. Oracle, DB2, and SQL Server would continue to compete effectively in things like scalability and performance, if for no other reason than they had a ten to twenty year head start.
As Ken North pointed out in his introduction, the number of standards bodies had mushroomed in the past few years and there seemed to be a genuine turf war in the standards world. This was reinforced by a slide he showed of the standards and organizations involved in Web Services, which looked like a Venn diagram drawn by an overly-enthuiastic kindergartner.
Jim Melton agreed with this statement, noting that some standards organizations, such as ANSI, ISO, and the W3C will survive, while others will play their part and then disappear. He also pointed out that, while newer standards organizations had been founded on the premise that standards needed to be developed quickly, they had been no more successful than the established organizations in developing standards in less than "geologic time."
Peter Coffee pointed out that customers benefitted from standards turf wars, since it meant that vendors were arguing over standard definitions of functionality, rather than proprietary ones. Furthermore, he felt that there was a real growth opportunity for companies willing to verify standards compliance and that customers were willing to pay for this.
A number of other topics were discussed in less detail. Rick Cattell said that Java would continue to evolve and that various expert groups were determining how to use Java with a variety of technologies. Peter Coffee noted that the best thing about today's security problems was that everybody in IT was now expected to be part of the solution. Coffee also noted that the market for software for mobile devices presented a huge opportunity, pointing out that this was probably the year in which more handheld devices like PDAs and cell phones would be shipped than PCs.
Audio and video from the discussion is available online on the Web Services Summit Web site.
Front Page | About | Contribute