IBM dismisses BEA productivity boast
My benchmark is better than yours
Last week BEA claimed its WebLogic Workshop is more than five times more efficient for development of web services. BEA posted an example human resources application on its dev2dev community web site the company claimed can be built in just five steps. BEA claimed it takes 28 steps to build the same service using WebSphere Application Developer.
IBM's Henrik Hedegaard, VP of WebSphere for EMEA, told ComputerWire that IBM had not yet had a chance to verify BEA's claims against its own data, but was sure that the claims were unfounded: "It's easy to issue a press release," he said. He added he, "didn't believe that the [claims] were likely," but that IBM would be happy to submit WebSphere to any public benchmarks BEA challenged it to.
IBM and BEA are locked in a battle for supremacy in the application server market, with some analysts such as IDC saying that there is little, if anything, between them in worldwide market share. However in an exclusive interview with ComputerWire earlier this week, BEA's CEO, chairman and co-founder Alfred Chuang, said that analyst market share figures were, "a million miles from the truth", because IBM does not break out revenue figures for WebSphere, making the market share estimates little more than guesswork.
IBM's Hedegaard once again refuted those assertions, saying that while IBM does not break out revenue figures it has told analysts that growth from the WebSphere product was 43% in Q4 2001, 53% in Q1 2002, and 19% in Q2 2002. "Compare that to BEA's revenue growth and you can see who is winning this battle," said Hedegaard. In its second quarter BEA's sales were down 15.6% on the same period a year earlier. Its first quarter saw sales down 12.6% year on year.
Hedegaard also insisted that IBM would not issue counter-claims to BEA's productivity argument, because, "customers are not motivated by marketing claims. They know that if one application server was four or five times better than another then that company would own the market." Which is exactly what BEA's Chuang argues his company does.
However, IBM does engage in the ECperf application server benchmarking process, which measures the performance, scalability and price-performance of J2EE application servers. While BEA's claim is of a productivity advantage in its development environment rather than a performance lead for its application server, the ECperf benchmarks currently show Oracle's application server to be performance leader running on Sun hardware and also price-performance leader running on HP, while IBM's WebSphere came third in performance running on its own hardware and seventh on price-performance. BEA does not appear in the top 10 for performance, and comes fifth for price-performance running on Dell hardware.
However IBM's Hedegaard said that even the ECperf benchmarks are not a good guide to application server superiority, because the performance and price-performance leaders alternate regularly between the different vendors. "Customers aren't looking for a vendor that's ahead in benchmarks for three months," he said, "they're looking for a vendor they can partner with and who will offer them the required level of support and quality of service."