Forum Discussion

Francisco_1_303's avatar
Francisco_1_303
Icon for Nimbostratus rankNimbostratus
Aug 22, 2013

define persistency with irule on gtm

i have few web servers that are behind gtm. the wide ip is using topology as the service is global and the pool level is using global availabilty. the service requires stickness persistency so hence why we using global availabilty but the problem is the server in the pool with the lowest order is getting hammered. I was looking at irule to define something on the gtm to utilize the rest of the server within the pool like a rounb robin fashion and still maintain persistency. global availabilty doesnt really achieve that as it always return the same server in the pool with lowest order. Is there anything available on the gtm to achieve what i want? i want to be able to utilize all servers and keep my persistent.

 

6 Replies

  • Francisco -

     

    What are you using to do the local load balancing at each datacenter? Those types of decisions would be made there. If you're not doing any local pools, let me know, and I will think about how to do it at the global level.

     

    Thanks, Jeff

     

    • Francisco_1_303's avatar
      Francisco_1_303
      Icon for Nimbostratus rankNimbostratus
      Hi, so the severs are behind the gtm's (4 gtms in a sync group in each datacenters global). no ltm involvement just gtms. the pools contain members from each datacenters and load balacing is global availability. the wide ip is define as topology.
  • Francisco -

     

    What are you using to do the local load balancing at each datacenter? Those types of decisions would be made there. If you're not doing any local pools, let me know, and I will think about how to do it at the global level.

     

    Thanks, Jeff

     

    • Francisco_1_303's avatar
      Francisco_1_303
      Icon for Nimbostratus rankNimbostratus
      Hi, so the severs are behind the gtm's (4 gtms in a sync group in each datacenters global). no ltm involvement just gtms. the pools contain members from each datacenters and load balacing is global availability. the wide ip is define as topology.
  • how many pools and how many pool members per pool? Can you clarify your design goals?
  • Hi, There is 7 pools, about 4 memebers each. The servers within the pools are used as web proxy for internet browsing. my goal is to utilize all servers within the pool and maintain persistency as the web traffic going through these boxes are web base transactions that requires persistency. global availability is not working. I know if i have an LTM i could achieve my goal easy. :)