Cisco ip sla options


(Lior) #1

Anyone able to share ip sla setup on cisco for monitoring gre service availability?
i am aware of the best practice published on the public kb, however sla seems to fail often, possibly due to local packet loss.
Any proven setup to share? timeout value? threshold? frequency?


(Mike Richard) #2

Take a look at the ‘delay down’ option.

Assuming you are using http tests, the most frequently they will run is every 60 seconds. The ‘delay down’ option allows you to delay a failure event for a given number of seconds. This would allow for multiple failures before triggering an event. You may want to ensure you are using GRE keepalives as well to enable failover during a hard outage, otherwise IP SLA will wait for the specified number of seconds to fail.

An example of how to apply this:

track 1 ip sla 1
delay down 120

(Sumanth Malempaty) #3

GRE.pdf (951.4 KB)


Just seconding Mike and the attached doc has a configuration example.

If you use HTTP IPSLA alone - it takes 3 X 60 = 3 mins for the failover to start. Some companies are not comfortable with such a long failover time.

If the GRE router has the public IP on its interfaces, you can use a combo of GRE keepalives and HTTP IP SLA as suggested by Mike!

If your GRE routers are behind NAT, the GRE keepalives don’t work. You may want to pair the HTTP IPSLA with ICMP IPSLA to trigger a quicker failover in hard down situations.

Sumanth M

(Lior) #4

Is this document missing the track association with the route?