Forum Discussion

11 Replies

  • Amr_Ali Do you mean having an HA device much like you have HA LTMs? If that's a yes you really don't have to do that because GTMs are inherently redundant for DNS because of how the sync works between them for DNS configuration. You might need them to be HA because they run multiple functions such as GTM, LTM, AFM, and so on but you should try and keep GTM off of the same box as those other services if possible.

    • whisperer's avatar
      whisperer
      Icon for MVP rankMVP

      I agree with Paulius 100%. If the F5 BIG-IPs only running GTM as dedicated module, no need for HA! Use GTM sync group and publish both listener addresses for DNS... or setup DNS anycast. If in HA, you don't have BOTH units responding to DNS (normal setup), but if both are standalone and part of the same GTM sync group, then you still have resiliency and depend on iQuery between both GTM instances and DNS for failover. If running more modules... like LTM and/or ASM the conversation changes. Most clients with run dedicated smaller devices for GTM/DNS only, and a bigger device for all other ADC functions.

      • Amr_Ali's avatar
        Amr_Ali
        Icon for MVP rankMVP

        hello whisperer, I need to make HA between two GTM boxes at the same location ( datacenter ) not in different locations

  • So, I see the above comments and have something similar story now. 

     

    We are in the process of upgrading from the Guest GTM to a new Tenant GTM appliance. Our plan involves adding the new GTM to the cluster sync-group, performing a manual full synchronization, and monitoring its performance for a day. Once verified, we intend to gracefully take the old GTM offline. If all functions well with the new GTM, we will then remove the high availability configuration and proceed with cleaning up the old GTM. Any suggestions or recommendations on this approach?