Posted: 11 May 2018 11:22 EDT Last activity: 12 Oct 2018 14:18 EDT
Using bigip kind of AES URL in the Monitred node PDC landing page-need resoltion at the earliers
We have AES7.2 Application installed on top of Pega 7.2.2 which has 4 jvm app servers and 2 webservers.
Usually if the AES application is installed in one jvm then we can use the PRSOAPServlet url of the single jvm and update it to the PDC landing page of the monitored node.
Since we have multiple app and webserver, we thought of using loadbalance bigip soapservelet url, so that the soap messages from the monitored node will be sent through load balance big ip and the message will be received by any one of the AES 4 app jvms and though webserver. Let us know if that works, otherwise how do we implement in such a way the soap message from the monitored nodes will be sent to any one of the AES app server since we have multiple aes app servers.
***Edited by Moderator Marissa to update platform capability tags****
AES servers should be deployed behind a load balancer. Simplest configuration is to have one node to run agents and elastic search - and keep that node out of the load balancer - and two or more nodes to load balance incoming message traffic and user sessions.
The PDC service uses three nodes in this configuration (two web, one batch) to monitor over 100 systems and process 2 million messages per day.
AES 72 agents are defined to run on all nodes but only "do work" on the node specified in dynamic system setting "PegaAES", "AES/SECURITY/AGENTS/NODEID". Adjust it accordingly.
So if you our infra setup, we have totally 2 webservers and 2 app servers(2 +2 jvms), so what your suggesting is have loadbalancer big ip url in front of 2 webservers that will route the traffic only to 3 jvms and 1 jvm can be purposed for batch processing ?
Web servers do not add very much to Pega. In theory one could use web servers to offload serving static content, but I've never seen it done, and static content overhead is pretty light. Some used web servers to offload https but load balancers can do that too - and modern hardware handles https pretty well. People often use web servers out of habit - plus they are part of the application server vendor reference deployment architecture. They're not going to do you harm once properly configured but I dont think they add value typically
So as per your comments, we can just a bigip loadbalancer in front of app servers rite... and have 1 jvm act to takecare of batch agent.remaining jvms as real time to handle request from monitored node rite..