cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
489
Views
0
Helpful
2
Replies

EIGRP SRRT anomaly (or not)

8r-murphy
Level 1
Level 1

I have several routers using VRF-lite to separate the layer 3 routing tables between two type of services. Each of these routers runs EIGRP in both VRF instances. The logical topology is the same as the physical. When I run "show ip eigrp vrf VOICE nei" on any router, the displayed SRTTs are all 1 ms. When I run "show ip eigrp vrf VISITOR nei" the displayed SRRTs are all 1 ms, except for one VLAN that carries a point-to-point link between two routers. This one shows an SRTT of 1587 ms. This uses the same physical connection that shows a 1 ms SRTT for the VOICE vrf. Why would these SRTT values be so different between the same routers over the same physical link?

 

Richard

 

1 Accepted Solution

Accepted Solutions

Peter Paluch
Cisco Employee
Cisco Employee

Hi Richard,

The SRTT is a moving average computed whenever a reliable (that is, to-be-acknowledged) packet is sent to a neighbor - Update, Query, Reply, SIA-Query, or SIA-Reply. The time difference between sending one of these packets and receiving an acknowledgment is used to compute the SRTT.

It may be that during the time when your router was sending packets to its neighbor in the VRF VISITOR, the neighbor was busy or there was some congestion on the link, causing the acknowledgment to arrive relatively late and causing the SRTT to be quite high. In a stable network, the reliable packets as mentioned above are not sent because there is no routing information to be updated, and so the SRTT stays on the old value. I believe that if you "jogged" the network a little and perhaps created a dummy loopback network advertised in the VRF and did shut/no shut on that loopback a couple of times, causing your router to send reliable packets to its neighbors, the SRTT would go down to 1ms fairly quickly.

As long as your EIGRP peerings are stable and the network does not show any significant latencies, I would not worry too much about this - but if that is possible, I suggest you do the little trick with the loopback :)

Best regards,
Peter

View solution in original post

2 Replies 2

Peter Paluch
Cisco Employee
Cisco Employee

Hi Richard,

The SRTT is a moving average computed whenever a reliable (that is, to-be-acknowledged) packet is sent to a neighbor - Update, Query, Reply, SIA-Query, or SIA-Reply. The time difference between sending one of these packets and receiving an acknowledgment is used to compute the SRTT.

It may be that during the time when your router was sending packets to its neighbor in the VRF VISITOR, the neighbor was busy or there was some congestion on the link, causing the acknowledgment to arrive relatively late and causing the SRTT to be quite high. In a stable network, the reliable packets as mentioned above are not sent because there is no routing information to be updated, and so the SRTT stays on the old value. I believe that if you "jogged" the network a little and perhaps created a dummy loopback network advertised in the VRF and did shut/no shut on that loopback a couple of times, causing your router to send reliable packets to its neighbors, the SRTT would go down to 1ms fairly quickly.

As long as your EIGRP peerings are stable and the network does not show any significant latencies, I would not worry too much about this - but if that is possible, I suggest you do the little trick with the loopback :)

Best regards,
Peter

Peter

You are right..this must have been some fluke because I did what you suggested and the SRTT dropped down to 1 ms in the cases where it had been 1597 ms.

 

Thanks.

Richard

 

Review Cisco Networking for a $25 gift card