This makes it so that the error model doesn't create loss bursts that are longer than the burst parameter. This should prevent idle timeouts when loss bursts are too long. It will however lower the overall loss rate below what is configured, because we're not adjusting the drop rate upward to compensate for breaking loss bursts (by forcing a forward).
This makes it so that the error model doesn't create loss bursts that are longer than the burst parameter. This should prevent idle timeouts when loss bursts are too long. It will however lower the overall loss rate below what is configured, because we're not adjusting the drop rate upward to compensate for breaking loss bursts (by forcing a forward).