Protocol promises faster Web services

A proposed method for sharing information between systems linked on the Internet promises to speed collaborative applications by up to 10 times the current rates.

The protocol, developed by Jonghun Park, a professor at Pennsylvania State University's School of Information Sciences and Technology, is based on an algorithm that lets it use parallel instead of serial methods to process requests. Such a method boosts the efficiency of how resources are shared over the Internet. The new protocol is called Order-based Deadlock Prevention Protocol with Parallel Requests.

For many years computer scientists have been proposing protocols to improve the efficiency of distributed computing systems, but Park asserts that his method works with greater efficiency for time-critical applications. The current protocol is generally known as the Order-based Deadlock Prevention Protocol, according to Park.

While the performance results of the new protocol have impressed analysts, they note that other more critical issues holding back collaborative Internet computing and Web services need to be addressed first.

"Web services is currently held up--in my opinion--by things like security and reliability," said Stephen O'Grady, an analyst at RedMonk. Once those concerns are addressed, people will "turn their attention to something like this protocol, which would offer incremental improvements in performance."

Although the existing protocol provides sufficient speed for today's databases linked to the Net for Web services and high-powered grid computing, Park said that current methods will face a number of challenges as Internet-connected computing becomes more pervasive.

For one, he said, the current protocol will lack the power to allow efficient coordination between Internet applications as they grow larger in scale, according to the researcher. Two other problems likely to become even bigger hindrances as Web services and Internet collaborative computing become widespread are livelock and deadlock. Livelock occurs when two or more processes keep shifting their requests in response to the changes occurring in the other--with neither process succeeding in carrying out any productive work. Deadlock occurs when two or more processes get bogged down waiting for the other to make a move.

In other words, instead of concurrent applications collaborating, they will vie for resources or just freeze while waiting for the other to take a lead.

"Better coordination will be required to meet that demand, and this protocol provides that," said Park, who presented his research this week at the Institute of Electrical and Electronics Engineers' Symposium on Applications and the Internet in Orlando, Fla. His paper, titled "A Scalable Protocol for Deadlock and Livelock Free Co-Allocation of Resources in Internet Computing," has not been published yet.

"The proposed protocol is free from deadlock and livelock, and seeks to effectively exploit the available alternative resource co-allocation schemes through parallelization of requests for required resources," Park wrote.

Park said that he will seek to commercialize the next generation of his protocol that he has been fine-tuning over the past year. Those refinements, he said, make the protocol less theoretical and more appropriate for real-world use. In September, Park will present the latest generation of the protocol at a symposium in Australia.