Routers/Switches /

Cisco to enhance IOS for reliable networking

By Stephen Lawson
IDG News Service, 05/14/02

Kicking off an initiative to provide for networks that just keep on going, Cisco Tuesday outlined software features it will add to its IOS software over the course of this year to make network failures or updates invisible to end users.

The set of features, collectively called Cisco GRIP [Globally Resilient Internet Protocol], are designed to let routers and the links between them continue operating in case of failures or planned upgrades to either hardware or software.

As IP networks increasingly are used for time-critical applications such as Web services, data storage, voice calls and video streams, corporations and other users need to be able to rely on them at all times, Cisco President and CEO John Chambers said last week in a keynote address at NetWorld+Interop 2002 Las Vegas. The Tuesday announcement follows Chambers' emphasis on "resiliency," a concept other vendors also were emphasizing at the show.

The Globally Resilient IP features include the following:

Nonstop Forwarding (NSF), which is designed to maintain route state information between two processors in a router so the standby processor can take over without disrupting the network or causing any loss of packets. It will be available in June, according to Cisco.

Stateful Switchover (SSO), which maintains link-layer state information on ATM, Ethernet, Frame Relay and other network links so packet forwarding can continue in case of a route processor failure. It will be available in June.

IP Event Dampening, in which software in a router can detect a "flap" (in which a link in the network fails and recovers quickly and repeatedly) and keep the router from using that link until it is stable again. It will be available in June.

Multi-protocol Label Switching Fast-Reroute Node Protection, which enables an MPLS tunnel to be rerouted around a failed node in milliseconds. It will be available in June.

Multicast Sub-Second Convergence, in which a multicast, such as a video or audio stream to many clients in an enterprise, can be rerouted in less than one second in case of a failure. It will be available in June.

Border Gateway Protocol Convergence Optimization, which can dramatically reduce the time it takes a router to recalculate the routes on a network. In tests, it allowed a router to converge a routing table of about 200,000 routes 40% faster than before, according to Charles Goldberg, a product line manager in Cisco's IOS group. It will be available in June.

Stateful IPSec and Stateful NAT [network address translation], which help a router at the edge of a network rebuild encrypted tunnels and private IP addresses quickly if one router fails and turns over its work to a standby platform. It will be available in the second half of 2002, according to Cisco.

Gateway Load Balancing Protocol, which allows enterprises to make use of a standby WAN connection even while the primary link is working. This could effectively let some enterprises double their WAN bandwidth. It will be available in the second half of 2002.

Cisco's new software features come as part of a wave of new capabilities from various vendors to make IP data networks more like the public switched telephone network, according to Hilary Mine, an analyst a Probe Research, in Cedar Knolls, N.J.

"The routing industry has been working pretty hard over the past two years ... turning the next generation of routing into a reality," Mine said.

"The router has become much more carrier-class, which is an economic thing as well as a reliability thing," she said. More resilient routers mean service providers don't just have a more reliable service but don't have to buy one backup for every router on the network.

Although Cisco is using some proprietary technologies in GRIP, the networking giant has a good track record of introducing new technologies to standards bodies and driving them to standardization, Mine said.

The success of Web services and of outsourced offerings such as Internet-based data storage will depend on reliable networks, Mine said, using as an example a small business that adopts Internet-based data storage.

"If they have a lot of problems accessing the storage, they're never going back," she said. "People will tend to blame the service . . . regardless of the real cause."

The IDG News Service is a Network World affiliate.