In this post, I will explain the Overload bit which is an important feature of the IS-IS routing protocol.
When a router that runs an IS-IS routing protocol has a resource issue (CPU, Memory), the device shouldn't receive network traffic. It shouldn't be a transit router between any two other nodes in the network.
Routing protocols have the capability to signal other routers for this. A router that runs routing protocol can send a warning to other routers to stop sending transit network traffic through them.
In OSPF, the ‘ Max-metric router LSA ‘ feature is used and works in a very similar way to the IS-IS Overload bit.
When a router that runs IS-IS Routing protocol sets the Overload bit in its LSP (Link State PDU), routers detect that the Overload bit is set thus they don't use this router for transit traffic but they use it for packets destined to the overloaded router's directly connected networks and IP prefixes.
What happens when a router has a system resource problem and doesn't set IS-IS Overload Bit?
Other routers would still send the traffic to this router, but since there is a resource issue, there will not be a place for incoming LSPs into Link State Database which would create database inconsistency. As you might know, in IS-IS, every router should have the exact same database at a particular IS-IS level.
Otherwise, this inconsistency creates routing loops. Thus, when a router runs out of system resources, it is important to alert the other routers with the Overload bit.
IS-IS and BGP Interaction to avoid traffic blackholing, How IS-IS Overload bit helps?
Let's look at the below topology.
Figure - IS-IS and BGP Interaction - Overload Bit helps to avoid traffic blackholing
In the above topology, Router B will be taken down for maintenance purposes. Before the maintenance window starts, Router B should alert other routers that they shouldn't send traffic anymore through it.
This is done with the IS-IS Overload bit. But what is BGP and IS-IS Interaction? How Overload bit will help to avoid traffic blackholing?
In the above topology, let's imagine that BGP is running between each and every router (Full Mesh IBGP).
When Router B sets Overload Bit, it doesn't receive any transit traffic any more. That's why everything is normal.
But when Router B comes up after the maintenance, if the Overload bit is cleared before the BGP convergence, traffic is blackholed at Router B.
Let me explain in this way. When BGP destination traffic, comes from Router A to Router B, Router B needs to do an IP destination-based lookup. Who is the destination? BGP prefix. Did BGP convergence? No!
That's; why, until BGP converges, Router B should stay away from the networking traffic. Overload Bit stays as set until BGP finishes its job, when BGP converges (With End of Rib Marker message or Keepalive it is understood that BGP is converged) IS-IS Overload Bit is cleared and Router B start receiving traffic again.