The hardest thing about large networks is getting back to normal quickly after problems and instability.
You already know that OSPF is a link-state protocol that lets other OSPF-enabled routers know about changes in the network’s structure. OSPF runs the SPF algorithm at this point.
The issue comes up when network instability causes LSAs to arrive too quickly; in this case, the SPF calculation needs to be run every time a new LSA appears (LSA-driven), which could lead to high CPU usage. This is where SPF throttling comes in.
SPF throttling in OSPF
One scheduling strategy that can be employed to postpone SPF calculations is OSPF throttling, which imposes an SPF-waiting time. The SPF will now wait for the LSAs to be added to the Link-State Database (LSDB) before initiating the SFP computation, even in the event of a large number of LSAs being received. The SPF calculation then begins once the SPF-waiting interval has passed. Because of this, more modifications are incorporated into a single computation. This method has a negative effect: it increases the convergence time. However, using this timer naturally boosts OSPF efficiency.
In a perfect world, the wait interval would be long while the network is unreliable and short when everything is running smoothly.
Cisco implemented an exponential back-off technique in IOS 12.2 that does this by adding three more timers to the waiting timer interval; We’ll give each of the three timers a different name so everyone can understand:
- SPF-start: The SPF-start variable represents the starting delay in the SPF schedule. The SPF-start timer is set to 5000 milliseconds by default at the moment.
- SPF-hold: The minimum amount of time that must pass between two consecutive SPF calculations is indicated by the SPF-hold value. The SPF-hold timer is set to 10000 milliseconds with the default setting.
- SPF-max: the maximum amount of time that must pass between two successive SPF calculations is indicated by the SPF-max. The SPF-maximum timer is set to 10000 milliseconds from the default setting.
The range of these timings, which are all given in milliseconds, is 1 to 600,000 milliseconds.
To set up OSPF SPF Throttling, use the timers throttle spf {SPF-start} {SPF-hold} {SPF-max} Process command for OSPF.
This is how it works:
When the first LSA comes in, the SPF calculation is put off until the number set by the SPF-start timer is reached. The SPF-hold timer’s value is used to set the SPF-waiting timer.
LSA(1) - (SPF-start=5000) (SPF-hold=1000) –> (SPF-waiting= SPF-hold)
During the time that the SPF-waiting timer is running, the router will increase the SPF-waiting interval to twice its value if it gets a greater number of LSAs. This means that the SPF-waiting interval will increase each time an event takes place during the hold-time window that is now active.
LSA(2) - (2*SPF-wait=20000)
LSA(3) - (2*SPF-wait=40000)
LSA(4) - (2*SPF-wait=80000)
When the present SPF-hold timeframe ends, the SPF calculation will be postponed; otherwise, the SPF-wait interval will keep increasing exponentially until it hits the SPF-max value. The SPF calculation will be carried out if this occurs.
After two times the SPF-max timer, the timers will be reset to their configured levels if no additional events are detected.
LSA throttling in OSPF
This function also has a dynamic scheduling feature that can lower the number of times LSAs are created when the network is unstable. The rate of LSA generation was limited for 5 seconds before this feature was added; this is the usual LSA-wait timer interval. In other words, the LSA could not be sent in milliseconds. Because of this, OSPF could not reach sub-second consensus. The idea behind it is like SPF throttling. It gives Adaptive Timers for making LSAs when the network isn’t stable.
To do this, OSPF LSA slowing adds three more timers to the waiting timer interval. All three timers are shown in the CLI as delay, so I’ll give them each a different name to make it easier to understand:
- LSA-start: “LSA-start” indicates the delay in the first schedule for LSA generation. There is a default value of 0 milliseconds for the LSA-start timer.
- LSA-hold: With LSA-hold, you can specify how long you must wait between two successive LSA computations. The SPF hold timer is initially set to 5000 milliseconds.
- LSA-max: The maximum amount of time that can pass between two successive LSA calculations is indicated by LSA-max. The maximum LSA timer is set to 5000 milliseconds by default.
Utilize the in order to configure OSPF LSA Throttling at timers throttle lsa all {LSA-start} {LSA-hold} {LSA-max} OSPF process command.
This is how it works:
The first LSA is sent right away after an event by default, and the LSA-wait timer is set to LSA-hold interval.
LSA(1) – (LSA-start=0) (LSA-hold=5000) –> (LSA-waiting= LSA-hold)
When more LSAs are required, they will be generated at the LSA-hold interval; in such a case, the LSA-wait interval is doubled.
LSA(2) – (2*LSA-wait=20000)
LSA(3) – (2*LSA-wait=40000)
LSA(4) – (2*LSA-wait=80000)
When the LSA-hold timer interval ends, all of the LSAs will be transmitted together if there are several events during the LSA-waiting interval. Until it hits the LSA-max value, the LSA-wait interval will keep increasing exponentially. The grouped LSAs will be created if this occurs.
All timers will be reset to their configured values if two times the LSA-max timer passes without any additional occurrences.
You must additionally configure the minimum interval for accepting the same LSA when using OSPF LSA Throttling. Let’s clarify: Dropping an LSA occurs if its arrival time is less than the interval. Cisco now suggests that the arrival interval match or be shorter than the LSA-hold timer interval of the throttle lsa all command.
To do so use the timers lsa arrival {milliseconds} command.
For lab purpose, we will use the following topology:
Example 1:
Let’s configure SPF throttling on both routers with the following parameters: 5 milliseconds initial delay, a 10 seconds hold timer, and 90 seconds of expiration.
In R1& R2:
!
router ospf 1
timers throttle spf 5 10000 90000
exit
!
Okay, the SPF throttling has been configured up on both routers as shown below:
R1#show ip ospf | include SPF
Initial SPF schedule delay 5 msecs
Minimum hold time between two consecutive SPFs 10000 msecs
Maximum wait time between two consecutive SPFs 90000 msecs
Incremental-SPF disabled
SPF algorithm last executed 00:14:42.056 ago
SPF algorithm executed 2 times
R2#show ip ospf | include SPF
Initial SPF schedule delay 5 msecs
Minimum hold time between two consecutive SPFs 10000 msecs
Maximum wait time between two consecutive SPFs 90000 msecs
Incremental-SPF disabled
SPF algorithm last executed 00:15:05.508 ago
SPF algorithm executed 2 times
Let’s put the configuration through test. Let’s flap the loopback 0 interface in R2 so that we may accomplish this.
