Cisco has announced an alternative to the Web-services protocol SOAP — and made it open source. Cisco says its Etch messaging protocol will be more efficient than the SOAP standard and the company will release the source code.
A beta version of Etch will be launched mid-year, as part of version 2.5 of the Cisco Unified Application Environment (CUAE). Within that environment, it will replace the W3C (World Wide Web Consortium) standard, SOAP, says Cisco, providing a more lightweight and faster interface, suitable for real-time interaction.
The SOAP standard is not well supported across different platforms, with different vendors supporting different subsets, said Louis Marascio, director of engineering for the CUAE platform at Cisco, according to a report in CIO.com. Etch will be 100 per cent compatible on all supported platforms, he said.
SOAP interfaces are defined in a complex file in the W3C standard WSDL (Web services description language), but Etch will use Cisco's own interface definition language. This makes it more efficient, claims Cisco, which quoted its own test bed for 900 SOAP calls a second that could manage 50,000 Etch messages a second, or 15,000 full two-way transactions based on Etch.
Etch will initially support C# and Java, with Ruby, Python and C planned, and the software will be open source so developers can incorporate it into products. The licence will probably not be GPL, but a less restrictive licence such as the ones used by Apache or Mozilla, reports CIO. Cisco is also considering handing over Etch to the Internet Engineering Task Force (IETF) standards body.
"Cisco's open sourcing of Etch probably has less to do with any corporate love for open source than with a realisation that the most viable way to take on an incumbent in an established software market is with open source," commented CNET News.com blogger Matt Asay. "Proprietary software is a way to guard one's position. Open source is a way to create a new position."
Others are sceptical. "Etch is really just adding more stuff to be developed, tested, deployed, managed, maintained and integrated, yet it doesn't actually solve any new problems or solve any old problems better than what already exists," said middleware expert and IEEE senior member, Steve Vinoski, in a blog post.
Vinoski said faster performance is not enough to recommend a new protocol, when the bottleneck in a distributed system probably lies elsewhere, and other issues beyond speed will affect the working of a system.