Forum Discussion

Bill_Mayo's avatar
Bill_Mayo
Icon for Nimbostratus rankNimbostratus
Mar 08, 2017

Best Practice for GTM/LTM Monitors

I am having some trouble understanding the interaction of the health monitors on a GTM vs the associated LTM and the best way to be sure that a node/service is up without sending it unnecessary traffic. Our scenario is this: We have 2 ISP's. There is a virtual server created for a given service (e.g. a web server) for each ISP. Each of these virtual servers points to the same LTM pool. The 2 virtual servers are then put in a GTM pool associated with the DNS record. I had thought that having an http(s) monitor on the GTM pool and the LTM pool was the correct way to go. Based on reality and some further research, I am no longer sure that is accurate. Is it true that in this scenario, the nodes are getting health monitor requests from 3 locations (virtual server 1, virtual server 2, and LTM pool)? If so, what is a best practice way to configure health monitoring here?

 

Note: As you can infer from my usage of "GTM", we are on version 11.

 

2 Replies

  • Let the LTM monitor the pool members. GTM will obtain the status of the pool members, assuming proper synchronization.

     

    In my opinion, there isn't a real need for GTM to send monitors to LTM's pool members. I have utilized GTM monitors only when there are non-F5 devices like servers or non-F5 load balancer.

     

    • michaelvoight's avatar
      michaelvoight
      Icon for Nimbostratus rankNimbostratus

      Correct, for this to happen the GTM must have bigip monitor configured for the Servers (which are LTMs). Then the use the LTM VIPs as the vservers on the GTM. The LTM will report the status of the VIPs to the GTM (to determine which vservers are up)  The vserver makes up the GTM pool.  WideIP passes out address based on the pool member addresses, which are actually the LTM VIP addresses