what is the pros and cons of no keepalive command in bgp and framerelay?

[huangzhy]

In bgp ,if i use no keepalive ,the hold time will not expire ,and the bgp session will always be consider alive.And in frame-rekay config,use the no keepalive to disable the LMI keepalive packets,but i don't know in which situation should i use the command?what is the pros and cons of no keepalve command?

[thisisshanky]

You wont be able send data across the Frame-relay link, if you disable keepalive, because LMI will get disabled. Use of no keepalive in Framerelay disables only frame-relay ukeepalive messages.

BGP has its own keepalive messages, which can be enabled or disabled. Both have same syntax(command wise) but are for their own purposes. Both shouldnt be considered the same.

[doug.lockwood]

I have used the Frame relay "no keepalive" command in two applications.

Several ATM Mux vendors had boxes that did not support LMI but did allow Frame relay. An example would be the ADC AAC3, and the Fore cellpath 90.

The second use is to run frame relay across a point to point t-1

In that configuration, the overhead of the frame-relay switching does not add any value. I was using this for traffic shaping and fragmentation.

Doug

[huangzhy]

If I turn off the LMI,how the router know the pvc status? And how the keepalive message is sent?

[doug.lockwood]

If the frame-relay PVC is over a point-to-point t-1, the only way the PVC will go down is if the physical t-1 goes down. In this case the keep-alive’s and the PVC status are redundant. Think of Frame-relay in this configuration as an HDLC protocol that supports VLAN tagging.

In a (virtual) switched Frame relay network(99% of frame relay implementations), the loss of LMI and keep-alive’s are undesirable. The only time LMI would be turned of is if the frame relay switches did not support it. In this rare case, a dynamic layer 3 protocol is required to detect PVC failures and provide keep-alive’s.