Configure Hardware Load Balancer at HTTP Server Layer

Yesterday I discussed about Overview of Load balancer and various types of load balancer as you can configure load balancer at various levels like
DNS Layer Load balancing
HTTP Server level Hardware Load balancing(Discussed today)
Jserv level load balancing (I’ll discuss after HTTP level load balancing)

Today in this post I am discussing only at HTTP Server level Hardware Load balancer.
Before configuring Load balancer in Oracle Apps , you should be familiar with following technologies
1. Cloning
Check my previous post to know about cloning
Cloning Basics
Cloning Part I
Cloning Part II
Metalink Note # 230672.1 Cloning Oracle Applications Release 11i with Rapid Clone

2. Load Balancer Basics
Check my previous post for Over of Load balancer

3. Autoconfig
Check About Autoconfig on My Site
Metalink Note 165195.1 Using AutoConfig to Manage System Configurations with Oracle Applications 11i

Broad Level steps for configuring Hardware Load balancer at HTTP Server layer
A. Configure Hardware Load Balancer for Session Persistence
B. Apply Prereq. Patch (If you are not on 11.5.10 or higher)
C. Add additional Web Node using Rapid Clone
D. Configure Load balancer on Node1
E. Test Load balancer via Node1
F. Configure Load balancer on Node2
G. Test Load balancer via Node2
H. Test Load balancer is sending requests to both nodes when both Web Nodes are Up

Yesterday I posted that you need extra attention while configuring Load balancer at HTTP layer is about
Session Persistence – Hardware level Load Balancer should be configured for session persistence also called as stickness or sticky bit for Oracle Apps 11i . If this is not set & lets assume client1 first request went to Node1 after authentication if next request for client1 goes to Node2 who is not aware of client1 ‘s previous request will throw authentication again. Don’t worry about how to set it, ask vendor or Unix Administrator to set sticky bit or session persistence .

Now you have broad overview of configuring Load balancer at HTTP layer, I’ll discuss in detail why you need to apply above patches & what all step by step configuration and parameters which needs to be changed and reasoning behind those changes . Till then check for Cloning, Autoconfig.

If you are preparing for Apps DBA Interview’s and want to learn Apps via these questions then check sample questions with answers at my Site there are around 16 pages with 5-6 questions on each page . Do let me know if you want to discuss any Interview Question which you encountered or want to share with budding Apps DBA’s then mail me at

Happy Learning

Share This Post with Your Friends over Social Media!

About the Author Atul Kumar

Oracle ACE, Author, Speaker and Founder of K21 Technologies & K21 Academy : Specialising in Design, Implement, and Trainings.

follow me on:

Leave a Comment:

Hank says May 10, 2007

Hi, Atul,

I have a shared APPL TOP APPS structure of two nodes, we are using DNS F5BigIP load balancer(it seems to be the same as HTTP LB), but I found the two nodes’ Jserv Properties file has the same ports range defined, say, port=16750-16759(we do not use the conventional default port 8001/8007). If we are using Jserv LB, should we allocate the ports into two nodes? We have a F5 server using DNS LB, can you tell me the advantages and drawbacks of DNS and Jserv LBs?

We use session persistent(as most session will take long time, keeping them active over 2 hours) I found DB tiers are not load balanced well, one is pretty heat. I have a Plug-ins used with my grid control to monitoring the F5, but how can I figure out if LB works well or not?

Thanks a lot,


Atul Kumar says May 10, 2007

Hardware loadbalancer are always better and preferred over software loadbalancer as HW LB is stable and provide more configurations.

Your question about loadbalancer is not clear .

Do you want to check if LB is working or not meaning its transfering requests to both servers or not ?
If you wish to check this , check access_log on both servers and you should get hits on both web servers configured in pool of load balancer.

Hank says May 10, 2007


Thanks a lot for the information.
I am using the shared top structure, so the access log from two nodes merged, I did find some seesions from some clients, but it hard to to tell which apps node were hit.


Atul Kumar says May 10, 2007

Have you modified access_log location in httpd.conf files from two nodes to same location ?

You can check for processes “ps -ef | grep httpd” to check if new connections are coming to both servers or not, if this is one time test. shutdown first middle tier & test and then second middle tier

sree says July 12, 2007

Hi atul

i am using two nodes for oracle applications (forms,apache).now i want to configure oracle application with two url’s coz i am not using any bigip .
1)can i confgure two url’s in oracle applications for two nodes?
2) how can i loadbalnce without any bigip ?

please suugest us and give me configuration how to setup loadbalancer without using thirdparty loadbalancer or any native method in oracle applications for loadbalncer?

Thanks a lot


Atul Kumar says July 12, 2007


Yes you can configure two url’s for two nodes both pointing to single instance using profile options and context file.

2) how can i loadbalnce without any bigip ?

Use any other supported software load balancer, one which I am aware is Oracle WebCache as load balancer in front of E-Business Suite

Sandeep says June 24, 2008

Hi Atul ,

Which H/ware balancer have you used . can u pls recommend some good h/ware load balancer which can be used with oracle apps ?

Atul says June 24, 2008

Sandeep We use F5 but you can ask hardware vendors for others

ramyaksharma says June 17, 2011

Is there any way from client side to ensure that we always connect to same server behind the load balancer.
I tried using IP address instead of domain name to do so but failed to get the contents from server.

Add Your Reply