Loadbalancing / F5 and how to determine JDE / BSSV instance unavailable?

JDE0101

Active Member
Hi,

We are using BIG IP to loadbalance 4 instances of JDE on 5RHELinux WebServers. Thats 16 instances of JDE for around 1000 users. This bit works fine!

My question is how can we use BIG IP and what we know about JDE to "detect" that an instance is down and avoid routing new sessions to this instance?

The question is valid for loadbalancing the 20 JVMs and also 8 BSSV instances we use (we'd like to implement the same solution). Same principal in that we accept when an instance goes down users attached suffer but we want to avoid sending new sessions..

Any gurys out there care to advise!?


E9.1 / SQL Server 2008 / TR 9.1.3 / WAS on RHEL Web / Windows 2008 Logic/Ent/UBE
 
This isn't a 'robust' answer, but it works for us. We point the F5 at the cluster address on our 3 servers (We use OAS). Every 30 seconds or so we try to get a login screen, if a 200 (OK) is returned by the server, then we assume that everything is OK.

We let OAS decide how to allocate the logins to a particular instance.

Tom
 
WAS ND clustering should take care of the load balancing for you. We also use F5 with WAS ND 7 on Windows 2008-R2.
The F5 balances the HTTP web servers and WAS clustering balances the JVMs. If a JVM instances is off line, WAS the WAS plug-in will not route sessions to it. If an HTTP web server is not responding, the F5 will not route sessions to it. Works great for us. We, however, are not currently using BSSV so not too sure how it would behave.
 
I know this is an old thread but I'm hoping you can shed some light on how you implemented your solution to loadblance the BSSV instances?

I am currently I am trying to loadbalance using Citrix Netscaler - I have a HTTP Virtual server with 2 SSL Services on a roundrobin. The services show as down.

Any insight you can provide would be a great help.
 
Back
Top