Once Microsoft Corp., IBM Corp., and Sun Microsystems climbed on board, the Web services bandwagon started getting crowded. Lately, it's hard to find a software company that's not touting Web services as the next great thing in enterprise computing. Web services' promise of publishing applications on the Web for users of many types of devices and users to use is appealing. But, businesses should look before they leap on the bandwagon, said Thomas Kelly, CEO of SplitFire Technologies, an Ann Arbor, MI-based technology education and courseware development company.
"After all, Web services is a brand new, untested technology that will require extensive knowledge on the part of your developers," said Kelly. "And it could potentially fail."
Even with the risks, however, some companies could benefit from adopting a Web services model today. Kelly offers some guidelines for deciding if Web services are a good fit and insights into the challenges facing this nascent technology in part one of a two-part searchWindowsManageability series. In part two, he'll discuss the dangers of falling for the computer industry's hype about new technologies.
|sWM:||Should any business using the Web adopt Web services today?|
Web services sound great, but there's no reason for companies to scrap systems that
| are inexpensive and work well to use this technology unless there is a definitive reason. Not every business needs Web services. Today, most companies use the Web in this way: they have a standard thin HTML client contacting a J2EE container with servlets, Java server pages and enterprise Java beans. This is cheap and has a strong track record. It works well in small, midsized, and large company networks. If you don't require anything more than that, there's no sense in using Web services.
|sWM:||Which types of businesses are good candidates for Web services?|
Good candidates are companies which have integration issues, which have mainframes, client-server systems, PDAs, and many types of devices and
operating systems and applications on their network. Web services make more
sense in this kind of environment. Usually, this happens in a large company;
but there are no blanket statements in the computer industry. There will
always be exceptions.
|sWM:||In an organization that could benefit from Web services, where should the technology be used first?|
Web services will be a strong intranet technology. Web services are going to be a mechanism by which I can invoke components without having to
restrict myself to a particular client format. To make this happen, however,
an organization will have to be able to control their own internal standards. They need to be able to build components that service requests and have some level of freedom in terms of the language that I select to develop these components. Many businesses will use Java, I think, to build components of Web services on their intranets. Java is such a strong programming language. It's easy. There are a lot of tools available. It's got industry acceptance.
|sWM:||What's missing from the Internet-based Web services picture today?|
UDDI and open Internet, the whole concept of Web services description languages with dynamic discovery, isn't feasible. It sounds good on paper to say: "I'm going to discover an organization out on the Web. I'm going to be able formulate some sort of an indication and receive data back and work
with that data in a meaningful way." Yet, without a standardized interface,
how exactly do people plan to develop this kind of application? This is the
big unheard objection.
|sWM:||What is needed, then, to make Web services on the Internet feasible?|
I don't believe that using the Internet for Web services is really feasible in the absence of a formulized standard interface that numerous companies will be able to agree on. For example, all banks would have to agree on an interface. Then Web services for the wide-open Internet will be realistic. Thus far, however, competitive businesses have not been eager to agree on a standard interface. They don't think it's economically viable to agree on an industry-wide level. For Web services to happen on the worldwide Web, it would have to be economically viable for all the businesses in a particular industry to conform to a particular API. This would mean that competing businesses would be allowing customers to compare oranges to oranges, allowing the consumer to choose for themselves the most effective service at the cheapest cost. I don't see competing businesses rushing to do this. Besides, If they did, CORBA offers a much more sophisticated mechanism to achieve this sort of model.
FOR MORE INFORMATION