Packet Design Looks at BGP Roots

Packet Design to demo its BGP root-cause analysis technology, which it will release as an upgrade to Route Explorer, at Networld+Interop

May 3, 2004

3 Min Read

PALO ALTO, Calif. -- Technology that can help service providers quickly determine which single routing event is the root cause of millions of other events in their BGP (Border Gateway Protocol) networks has been developed by Packet Design.

Packet Design's BGP root-cause analysis technology will be available later this year as an upgrade to the company's Route Explorer appliance. When combined with Route Explorer's existing ability to provide full routing-path visibility and analysis of BGP networks, the new functionality will give network engineers unprecedented knowledge of these historically hard-to-diagnose networks.

The technology will be demonstrated at the Networld+Interop conference May 11-13, at Packet Design booth #651. BGP, used by virtually all service providers and many large enterprises to exchange routing information between interconnected autonomous systems, has posed major troubleshooting challenges because it is the most "chatty" of routing protocols. A minor connectivity change can produce hundreds or thousands - even millions - of updates following a significant peering loss.

"Today's networks are so large and complex that, while gigahertz processors and terabyte disks can capture and record the huge numbers of BGP events, making sense of these events in real time has been computationally unfeasible," said Jeff Raice, Packet Design's executive vice president of marketing and business development. "Advanced statistical techniques exist for pinpointing root causes from large volumes of data points, but until now it hasn't been possible to deduce the source of a problem in a timeframe that would be practical for resolving it."

Packet Design's technical team has leveraged its routing expertise to develop two new capabilities: a statistical algorithm that extracts the large-scale structure of BGP event streams to determine the root cause of those events, and visualization techniques to display that structure in operationally meaningful ways. The new algorithm provides the computational efficiency to examine millions of BGP events in real-time and identify the single or multiple events that triggered them, even if those events are several "hops" away from the user's network. It works by exploiting a real-time topological view of the network to analyze the "tree structure" of BGP events, thereby isolating individual sets of events resulting from a common source. This technique reduces the massive computational requirement to a manageable level, potentially narrowing down thousands of events to a single Syslog message.

"Tests have shown that our technology can keep up with a BGP event stream from a typical tier-one provider's entire backbone mesh using a 1-gigahertz CPU," Raice said. "When the algorithms are implemented in Route Explorer later this year, network engineers will be able to find and correct BGP problems much faster than ever before. They will quickly be able to see what happened, how it happened and how the network is affected. And they'll have real-time network topology views to help them rapidly diagnose and fix problems such as misconfigured community tags, policy filters with unintended consequences, and unexpected or unwanted backup paths."

Packet Design Inc.

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like