UCSC-CRL-00-20: Avoiding Routing Instability During Graceful Shutdown of OSPF

Aman Shaikhy, Rohit Dubez, Anujan Varmay
12/20/2000 09:00 AM
Computer Engineering
Many recent router architectures decouple the routing engine from the forwarding engine, so that packet forwarding can continue even when the routing software is not active. This implies that one can avoid route aps that occur when the routing process goes down provided the forwarding engine remains active during that time period. Unfortunately, the current definitions of routing protocols like BGP, OSPF and IS-IS do not support this behavior. In this paper, we propose means of extending OSPF by adding a new capability called IBB (I'll Be Back) Capability to it, so that packet forwarding can continue for a certain time period when the OSPF process is down. IBB Capability can be used for avoiding route flaps that occur when the OSPF process is brought down to facilitate protocol software upgrade, operating system upgrade, router ID change, and AS and interface renumbering. We identify the challenges, one of which is the inability of the inactive router to adapt to routing changes, in making OSPF IBB Capable and propose solutions. We then describe a prototype implementation of IBB Capability we have developed using GateD. Using the prototype in an experimental setup, we demonstrate that the overhead of IBB Capability is modest compared to the benefit it offers and has good scaling behavior in terms of network size.