Message processors are affecting connection persistence on load balancer

My team created an API proxy that interfaces with a web service for BladeLogic. We have Apigee Private Edge setup here that has a very large infrastructure (multiple routers, messapge processors, ext and int,fronted by load balancers). There is also a load balancer in front of our BladeLogic app servers. The load balancer is having trouble forwarding requests to the correct app servers. Requests must go to specific app servers, based on the source IP address.

We think the issue is occuring because we have multiple message processors that are randomly processing the requests, and the load balancer is seeing those message processors as the source address instead of the true source IP from the client that sent the request to Apigee.

Is there a practical workaround or fix to this type of problem? Is it possible to send a request directly to a message processor? This would not be good from a high availability standpoint, but we weren't sure if there were any better solutions. Is it possible to limit certain proxies to only use one message processor? Or is there an altogether better solution than limiting the message processors?

Diagram of the setup described above:9386-bl-api.png

0 0 90
0 REPLIES 0