98 degree web

The latest sign that the Internet is shaking off the stigma of the dot.com bust--e-commerce figures are still going up. On the consumer side, e-commerce portal site BizRate says fourth quarter online sales were up 40% at $17.4 billion (for a yearly total of $47.98 billion), while the number of purchases grew 37%. Jupiter Research reckons that sales for November and December should beat its forecast of $13.1 billion.

And that's just B2C action--the US Commerce Department says that B2B is where the real money is, accounting for 90% of e-commerce in the US. On the global level, B2B e-commerce is already close to the trillion-dollar mark and Gartner expectsa that B2B online revenues could hit as high as $8.5 trillion by 2005. If it helps, Intel announced in December that it had processed $5 billion of its revenues and supplier purchases during the year--around 10% of its revenues--using RosettaNet technology standards.

For the uninformed, RosettaNet is an initiative from the IT, electronic components and semiconductor manufacturing sector to create and use industry-wide, open e-business process standards for supply chain partners. It uses XML to allow companies to do business with each other using the Web by bridging the differences between each company's individual proprietary network systems.

"Basically, these companies sat down and developed a dictionary of XML terms to represent what a price is, or what a catalog number is, so everyone's computers can talk to each other," explains Tom Burns, director of Intel's Asia-Pacific Internet Solutions group.

RosettaNet, which started in 1998, is also arguably one of the telecoms/IT industry's first serious attempts at Web services, a concept that has been around for some time but has been generating a bit of industry buzz of late, not least from the growing acceptance of open protocols and standards that are necessary to drive Web services. The Web service paradigm has also promised benefits such as cost-efficiency of enterprise network operations, a feature that has become a priority for service providers and their corporate customers alike.

But that's just for starters. The Web services concept has been a constantly evolving one. What began as an effort to standardize the way remote-procedure calls are defined and executed across a heterogeneous network has broadened into something much bigger that growing niche markets like content delivery and data storage have been hinting at for years. Now more and more industry players are taking notice, and taking action.

"In the late '90s," observes Vincent Lum, chief strategy officer for Datacraft Asia, "enterprises were buying lots of new IP equipment on the promise that it would increase productivity and profit. The hot issue now is, now that they have the equipment, how will they actually use it?"

Before going any further, it's worth clarifying what a Web service actually is. Superficially, of course, it is a service provided on the Web, but Intel's Burns says it's more helpful to think of it as the most basic level: two computers sharing data.

To do that, the computers have to be able to agree on a way to carry the data, but they also have to recognize what the data is--such as RosettaNet allowing computers to agree on what a catalog number is. Computers already do this to a degree via standardized markup languages like HTML, as well as proprietary solutions, but newer and more flexible standards like XML, SOAR UDDI and others are promising to

"Go to the next level, and you're looking at interoperability between platforms, where you get the benefit of legacy systems sharing data with less rework than via the common way of doing it in the past," Burns says.

Indeed, the problem of legacy systems is one of the factors driving the need for Web services. Enterprise networks have been built over multiple layers of disparate, often incompatible technologies--trying to develop, deploy, manage, and integrate applications across such networks is one huge collection of challenges. The apps themselves are built by different groups of people in different places with different technologies, many of them at a time when open standards were hard to come by. Integrating and managing incompatible applications takes time and money, as well as frequent IT infrastructure modifications.

That kind of effort also means lost revenue opportunities, Burns says.

Standards under construction

"There's a wealth of legacy systems out there that can be used to push forward new usages that provide value added services that are interoperable," he says. "Value added services are one of the few ways for businesses to keep value for the shareholders. Almost all of these changes have come from software."

The push to develop open standards has come from several standards consortia, including [W.sub.3]C (World Wide Web Consortium and OASIS (Organization for the Advancement of Structured Information Standards). Several such standards, such as XML, SOAR and WSDL, have been in commercial use for a couple of years.

Late last year, the WS-I (Web Services Interoperability Organization)--founded by IBM and Microsoft, whose members include Sun Microsystems, BEA Systems, Hewlett-Packard, Oracle and Intel--announced a number of new Web service specs aimed at security, including WS-Trust (for assessing trust relationships) and WS SecureConversation (for providing a secure contest for message exchanges), as well as four other specs covering security policies for Web services.

The WS-I also published its Basic Profile Working Draft, which explains how XML, SOAP, UDDI, WSDL and XML Schema can be used to develop interoperable Web services.

The fact that the draft only came out in October and the final profile will only be out early this year is a clear indicator not only of how far the open standards push has to go, but also of how long it's taken to get to this stage, says Intel's Burns. "The convergence of computing and communications has been a slow painful process over the last 15 years," he says. "We've come a long a way, and we know that because we've actually got Sun, IBM and Microsoft sitting in the same room now. Progress has come from the ability of the industry to do low-level benchmarks of cooperation."

Burns adds that while Web services already exist and can be deployed now, it's early days even with key standards in place. "This will take many, many years, it's not something that can be done overnight, of course. But everyone recognizes that it has to happen. The success of PCI and USB are proof that interoperability between any platform is not only achievable, but lucrative."

Rethinking the network

The Web services model is also being addressed from the infrastructure level--not just by companies like BEA Systems, which talks of "application infrastructure" when promoting its platform solutions, but also hardware vendors like IBM and Juniper Networks.

IBM is already an active player in Web services via its WebSphere platform, but the company's new "e-business on demand" strategy looks to address the heart of the matter by enabling deeper-level drivers such as grid computing--that is, allowing an overloaded processor on one computer to connect to other computers somewhere in the world with processing power to spare and passing on the extra work--and autonomic computing that allows such networked systems to be self-optimizing and self-healing.

"The idea is that we can reduce the complexity and the cost of connecting the user to the service," says Joseph Zisken, director of industry solutions for global telecoms at IBM.

The same sentiment is driving Juniper's MINT (Model for Integrated Network Transformation) strategy, which targets carriers looking for cost savings and new revenues (in other words, everyone). MINT leverages off some of the basic technologies and tenets of service creation that Unisphere Networks made its fortune on before being acquired by Juniper last year, and repackages them along with Juniper's strengths in routing and MPLS to offer a fairly comprehensive plan for carriers to integrate their services onto IP.

"We've spent a lot of time communicating the benefit of integration to our customers," says Juniper's Asia Pacific VP Adam Judd. "What we want to do now is conceptualize all that and transform the business model and the industry to where it needs to get to."

MINT breaks down into three layers--bit transport and connectivity, traffic segmentation by criteria such as app or user group, and wire-speed packet processing--plus policy and control for each of those layers that allows for service customization not only for the carrier, but for the end user as well.