Network switch

RIT researchers create programmable network switch to improve IP search time

A team of IT researchers from RIT has created a new programmable switch that dramatically reduces search time, helping to improve router performance by over 100%.

The team developed the switch with a new, longest prefix matching algorithm they named CuVPP. The algorithm uses packet batch processing and cache locality for both instructions and data, taking advantage of vector packet processing (VPP), and uses a cuckoo filter in the cache as a fast preselection mechanism. . According to researchers at RIT, CuVPP can achieve up to 4.5 million searches per second with live traffic, which is higher than other popular approaches.

For the work, the team received a Best Paper Award at IEEE Cluster 2020, an international conference in the field of cluster computing. The research was led by Minseok Kwon, professor of computer science; Mr. Mustafa Rafique, lecturer in computer science; Krishna Prasad Neupane, Ph.D. Computer and Information Science. student; and John Marshall, Engineer Emeritus at Cisco Systems and 1989 alumnus in Computer Engineering.

Researchers set out to improve the processing time of programmable software routers, which are more flexible at scale than hardware routers, but are currently not as fast.

“Programmability in network switches has become increasingly important as virtualization is introduced into networks and software-defined networks become more prevalent,” Kwon said. “Yet the performance of programmable network switches on basic hardware is a major concern, in light of the ever-increasing network speed and the size of the routing table. “

When Internet users attempt to visit a website, the router must be able to efficiently find the destination IP address and forward the data packets. This research challenge has been studied extensively because it can be a computationally expensive and timely process. To shorten the search time for an IP address, the RIT team decided to use the cuckoo filter technology.

“We can put the filter in a cache – a very fast, very small memory – so that we can pre-filter the IP addresses, instead of looking for an entire table,” Kwon said. “The challenge in building this is that you have to go through the source code and modify it, to make sure the filter is small and doesn’t negatively impact things. “

The researchers evaluated CuVPP with real network traffic using the IPv6 prefixes of the operational routers. They found that their switch improves search rates by up to 24% over baseline VPP, 51% over VPP with Bloom filters, and 128% over VPP with one trial.

“As the Internet continues to grow, technology to improve the efficiency of IPv6 is needed,” said Marshall, who works with Cisco Systems in North Carolina. “This algorithmic technique can really improve the performance of packet forwarding in software routers.”

CuVPP is an open source prototype and researchers hope to integrate it into IP research in the future. For more information on CuVPP, read the article.

Source link

Leave a Reply

Your email address will not be published.