OSPFv2 Shortest Path First throttling

Rapid triggering of SPF calculations with exponential back-off to offer the advantages of rapid convergence without sacrificing stability. As the delay increases, multiple topology changes can occur within a single SPF. This dampens network activity due to frequent topology changes.

This scheduling method starts with an initial value after which a configured delay time is followed. If a topology change event occurs the SPF is schedule after the time specified by the initial value, the device starts a timer for the time period specified by a configured hold time value. If no topology events occur during this hold time, the router returns to using the initial delay time.

If a topology event occurs during the hold time period, the next hold time period is recalculated to a value that is double the initial value. If no topology events occur during this extended hold time, the device resets to its initial value. If an event occurs during this extended hold time, the next hold time is doubled again. The doubling occurs as long as topology events occur during the calculated hold times until a configured maximum delay time value is reached or no event occurs (which resets the router to the initial hold time). The maximum value is then held until the hold time expires without a topology change event occurring. At any time that a hold time expires without a topology change event occurring, the router reverts to the initial hold value and begins the process all over again.

For example, if you set the initial delay timer to 100 milliseconds, the hold timer to 300 and the maximum hold timer to 2000 milliseconds, the following will occur:

If a topology change occurs the initial delay of 100 milliseconds will be observed. If a topology change occurs during the hold time of 300 milliseconds the hold time is doubled to 600 milliseconds. If a topology change event occurs during the 600 millisecond period, the hold time is doubled again to 1200 milliseconds. If a topology change event occurs during the 1200 millisecond period, the hold time is doubled to 2400 milliseconds. Because the maximum hold time is specified as 2000, the value will be held at 2000. This 2000 millisecond period will then repeat as long as topology events occur within the maximum 2000 millisecond hold time. When a maximum hold time expires without a topology event occurring, the router reverts to the initial delay time and the cycle repeats as described.

Therefore, longer SPF scheduling values can be used during network topology instability.