OSPF Fast Convergence Tools - Event Propagation

In continuation to my previous post about OSPF Fast Convergence – Event Detection, I will share my ideas about OSPF Event Propagation tool. Event propagation mainly covers LSA generation process. Once the event has been detected, LSA is generated to reflect the change. LSA is not generated immediately; OSPF waits 5000 msec until generating new LSA. This is only used for Router and Network LSA.

To get OSPF faster convergence(Read More About OSPF High Availability), LSA’s can be rate limited. This will increase OSPF convergence time.

Below is the command line to configure LSA throttling
router ospf 10
timers throttle lsa all (start) (hold) (max)
timers lsa arrival

start:[default = 0 ms] Default is to generate LSA immediately after receiving first trigger.
Recommended = 10 ms

hold : [default = 5000 ms] Small increments of 20ms should be considered if multiple link failures may occur on same router. This ensures if all failures have not been advertised by first regenerated LSA, following one will be promptly triggered.
Recommended = 100 ms

max : [default = 5000 ms] Default value may be kept unchanged. Prior to introduction of LSA throttle timer, bevahior was to wait MinLSInterval between distinct originations of any particular LSA with default MinLSInterval = 5s. Default value provides response equivalent to pre LSA-throttle.
Recommended = 5000ms

timer: [default 1000 ms] This controls minimum interval for accepting the same LSA. “Same LSA" is defined as an LSA instance containing same LSA ID, LSA type, and advertising router ID. If an instance of same LSA arrives sooner than interval that is set, LSA is dropped.
Recommended = 80 ms


Click Here To Read Rest Of The Post...

OSPF Fast Convergence Tools

What is Network convergence?
Time required to “Detect the Event”, “Propagate the Event”,” Process the Event” and “Update the Routing and Forwarding Information base” and after that routing the traffic to an alternate path during the outage of primary network path is called network convergence.

OSPF have various tool and techniques for fast convergence as below mentioned:-
• Event Detection
• Event Propagation
• Event Processing
• RIB Update

Read OSPF High Availability Techniques with SSO,NSF and NSR

I will be covering various OSPF Event Detection tools in this post:-
• Carrier Delay
• Bidirectional Forwarding Detection
• Physical Links
• IP Event Dampening
• Fast Hellos

1. Carrier Delay:- When the physical interface states changes, software must notify that change to the routing process. By default carrier delay is 2 seconds. So we can configure the carrier delay to zero for fast link detection by using “carrier-delay msec 0” under interface.
Note:-Sometimes if the flap interval is less than 2 seconds, during that time router doesn’t generate any logs for the same. This could also be the reason that syslog will not see any message.

2. Bidirectional Forwarding Detection

3. Physical links:- If you are using PoS interface, link failure detection is almost < 50ms. Also no need to configure BFD on PoS interface. If you are using Gigabit Ethernet interfaces, use Auto-Negotiation under physical interface as it helps to detect unidirectional failures. By default it is enabled and recommends not disabling it.

4. IP Event Dampening:- By using this tool we can mitigate the flapping links. The concept is same as we normally use in BGP. It simply tracks the flapping interface and applies penalty. Add that flapping interface in down state from routing protocol perspective if it exceeds the penalty threshold.

5. Fast Hellos:- OSPF fast hello packets mainly focus on the stability rather than the convergence. It is very CPU intensive process and sometimes leads to the high CPU problems also. It can be configured under interface by using “ip ospf dead-interval minimal hello-multiplier 3” command. Minimal keyword in the command sets dead interval to 1 second and hello-multiplier 3 means send 3 packets in one second.


Click Here To Read Rest Of The Post...

CISCO EIGRP DUAL Algorithm

Enhanced Interior Gateway Routing Protocol (EIGRP) is a advanced distance vector routing protocol proprietary to CISCO. Highly valued for its ease of deployment and fast convergence, EIGRP is used in many large Enterprise networks. EIGRP maintains all of the advantages of distance-vector protocols and having good features set for selecting loop free paths too.

EIGRP scales effectively in a well-designed network and provides extremely quick convergence times with minimal network traffic. EIGRP advantages include:
• Consumes low network resources as compared to OSPF
• Transmits only partial updates not the full routing table
• Rapid convergence times for changes in the network topology

Deep dive to get better understanding of Cisco EIGRP DUAL Algorithm

What is Reported Distance?
As depicted in Figure 1, A has three directly connected neighbors to reach E. The distance reported by A’s neigbors to reach E is known as Reported Distance. So A has three reported distance from it’s neighbors as below mentioned:-
• B can reach E with a cost of 10
• C can reach E with a cost of 10
• D can reach E with a cost of 30

What is Feasible Distance?
The total cost of each link to reach from A to E is known as feasible distance. As depicted in figure 2, A has three feasible distance to reach E with below mentioned cost.
• A can reach to E via B with cost of 20
• A can reach to E via C with cost of 25
• A can reach to E via D with cost of 45
The best out of the three feasible distances will become the successor. So as per the above output, B is nominated as the successor.

How to select the feasible successor or loop free alternate? After selecting successor, DUAL will look for the Reported Distance of the neighbors and check which one satisfies the Feasibility Condition; Reported Distance < Feasible Distance. As per the above two outputs we can conclude that
• C can reach E with a cost of 10(Reported Distance), so C reported distance (10) < feasible distance (20). This path is a loop free.
• D can reach E with a cost of 30(Reported Distance), so D reported distance (30) > feasible distance (20). This doesn’t satisfies the condition of selecting feasible successor (Reported Distance < Feasible Distance). So DUAL will mark this path as loop.

So now A has marked B as its successor and C has its feasible successor. Let’s assume now link between A and B is down as per figure 3. Now EIGRP will examine the available paths to E and declared C as a successor (best path to reach E) which was earlier selected as feasible successor (Loop Free Path). Now EIGRP will again look for its feasible successor and as per the Feasibility Condition, Reported Distance from D is 30 which is greater than the Feasible Distance. So EIGRP will consider D as loop path and didn’t qualify it for the feasible successor.

Now assume, link between A and C is down as shown in Figure 4. Now EIGRP will look for loop free path but unfortunately, as shown in Figure 3, no loop free path is available. However, A has a neighbor D might have a loop free path to E. So A will put E into Active State and query to D, D sends an reply to its query that it has a loop free path available to E. Once A will receive the reply from D, it beings start using that path for E.

Read More About Remembering PE-CE EIGRP

Read More About Eigrp adjacency issues with TLV


Click Here To Read Rest Of The Post...