0

I'm a software developer with limited knowledge of F5 GTMs.

We are building web applications that will run in Edge within a company network.

We have a set-up of two data centres & a GTM (F5) across the two. We use Kubernetes inside the DCs with Istio. We are building web applications that hold state in the DCs.

I'm told the GTM will allocate an IP for the data centre it decides to direct to with a TTL.

Once the TTL expires the browser will call the GTM again to get a new IP which the GTM could allocate to the other DC.

We were hoping to use sticky sessions & hold state in application memory but the GTM & it's TTL seem to make this unreliable. What I don't understand is I've worked at other places with F5 GTMs, active active deployments and sticky sessions. Is anyone aware of any way to get this to work with the GTM set-up we have? We could invest more and externalize the state but would prefer not to.

Person of earth
  • 71
  • 1
  • 8
  • GTM (traditional GSLB) provides A/AAAA responses to queries from LDNS servers. It decides which response to give based on its understanding of the local server load balancing state. Are you using any LSLB services in front of your web application besides (assuming?) kube proxy? More details would be helpful. – Jason Rahm Nov 20 '20 at 18:10
  • Hi, yes we have F5 LTMs inside each DC infront of the Istio gateways. I've been told preference is not to use i-rules – Person of earth Nov 23 '20 at 07:59

0 Answers0