Forum Discussion

jamesdris's avatar
jamesdris
Icon for Nimbostratus rankNimbostratus
Jul 11, 2017

monitor interval and timeout value

F5 standard says that a standard timeout value should be 3 times interval + 1 sec. One of the considerations for deciding on the interval is how busy the application could be. We don't want to bog the application down with health monitor traffic. But does it mean we need to wait for 3 times + 1 second to gauge the application response? Isn't it too high a response time for any application. Provided we don't want to bog the application down with too many health monitor polls, do we also need to wait for too long to judge the app's performance? How was this standard for a 3 times interval + 1 sec standard reached?

 

application delivery
LTM

1 Reply

Sort By
  • nathe's avatar
    nathe
    Icon for Cirrocumulus rankCirrocumulus
    Jul 11, 2017

    James, this is always a balance between ensuring the backend pool members are adequately monitored and how long it is reasonable before downing a pool member if it goes offline. Monitor too frequently to reduce time to down a pool member and you may not give the application enough time to respond to the monitors and, perhaps, unnecessarily be marked down - especially in busy times. Increase the frequency means it takes longer for the bigip to mark a pool member down and affect users. There isn't a quick answer to this one, it all depends on the app itself and its responsiveness. Don't forget the interval can be reduced, to say 1 or 2 seconds, or increased, to 15 or 30 seconds, so it does give flexibility.

     

    Hope this helps,

     

    N