Fast convergence and priority-based convergence can improve IS-IS network convergence. Fast convergence speeds up network convergence by fast calculating routes, while priority-based convergence sets different convergence priorities for routes to improve network convergence.
IS-IS fast convergence is an extended feature of IS-IS that is implemented to speed up the convergence of routes. Fast convergence includes the following:
Incremental SPF (I-SPF): recalculates only the routes of the changed nodes rather than all the nodes when the network topology changes. This speeds up the calculation of routes.
In ISO 10589, the SPF algorithm is used to calculate routes. When a node changes on the network, this algorithm is used to recalculate all routes. The calculation takes a long time and consumes too many CPU resources, affecting the convergence speed.
I-SPF improves this algorithm. Except for the first time, only changed nodes instead of all nodes are involved in calculation. The shortest path tree (SPT) generated is the same as that generated by the previous algorithm. This decreases CPU usage and speeds up network convergence.
Partial Route Calculation (PRC): calculates only the changed routes when the routes on the network change.
Similar to I-SPF, PRC calculates only the changed routes, but it does not calculate the shortest path. It updates routes based on the SPT calculated by I-SPF.
In route calculation, a leaf represents a route, and a node represents a router. If the SPT changes after I-SPF calculation, PRC processes all the leaves only on the changed node. If the SPT remains unchanged, PRC processes only the changed leaves. For example, if IS-IS is enabled on an interface of a node, the SPT calculated by I-SPF remains unchanged. PRC updates only the routes of this interface, consuming less CPU resources.
PRC working with I-SPF further improves the convergence performance of the network. It is an improvement of the original SPF algorithm.
Intelligent timer: applies to LSP generation and SPF calculation. The first timeout period of the intelligent timer is fixed. Before the intelligent timer expires, if an event that triggers the timer occurs, the next timeout period of the intelligent timer increases.
Although the route calculation algorithm is improved, the long interval for triggering route calculation affects the convergence speed. Frequent network changes also consume too many CPU resources. The SPF intelligent timer addresses both of these problems. In general, an IS-IS network is stable under normal conditions. The probability of the occurrence of many network changes is very minimal, and IS-IS does not calculate routes frequently. The period for triggering the route calculation is very short (milliseconds). If the topology of the network changes often, the intelligent timer increases the route calculation interval to avoid too much CPU consumption. The original mechanism uses a timer with uniform intervals, which makes fast convergence and low CPU consumption impossible to achieve.
The LSP generation intelligent timer is similar to the SPF intelligent timer. When the LSP generation intelligent timer expires, the system generates a new LSP based on the current topology. The LSP generation timer is designed as an intelligent timer to respond to emergencies (such as the interface is Up or Down) quickly and speed up the network convergence.
LSP fast flooding: speeds up the flooding of LSPs.
In most cases, when an IS-IS router receives new LSPs from other routers, it updates the LSPs in its LSDB and periodically floods the updated LSPs according to a timer.
LSP fast flooding speeds up LSDB synchronization because it allows a device to flood fewer LSPs than the specified number before route calculation when the device receives one or more new LSPs. This mechanism also speeds up network convergence.
Priority-based IS-IS convergence ensures that specific routes are converged first when a great number of routes need to be converged. You can assign a high convergence priority to routes for key services so that these routes are converged quickly. This reduces the impact of route convergence on key services. Different routes can be set with different convergence priorities so that important routes can be converged first. This improves network reliability.