Friday, April 18

Load Balancer

what is the difference between hardware load balancer and software loadbalancer and which loadbalancer we are using in weblogic?

Mukesh Negi 
HLB's are like Bigip F5, Cisco Ace, Oracle Traffic Director,
Software LB's are like apache, iis, iplanet. 
Depend on architecture combination of both or any one can be used.

Dilruwan Madubashitha 
The difference is the layer that it each works in OSI Reference Model, 
Hardware Load Balancer would either work in Network layer or transport layer.
Software LB will work in application layer.


Why we need to use software load balancer?
Although it is possible to use only the external load balancer's in the architecture, it's highly recommended to include the Web Server and the plug-in. The plug-in contains all the logic for the load balancing, session stickiness, and in particular the clustering fail-over with the transparent connection fail-over. More importantly, the plug-in is needed when setting up the workload management through work managers.
The biggest problem with load balancing is failure mode detection. detecting the failure correctly is the biggest issue, and the plugin gives you flexibility with that as well as session persistence etc. The plugin also gives you the ability to recover sessions in progress when there is a front end Apache to back end Weblogic .
Common design for us is a front end load balancer for the Apache (OHS) servers with the plugin supporting the clustering further on.


How the load balancer(Big IP) gets the requests from users?
How it redirects to Webserver(Apache)? What are the configurations required?

Maheshkumar Subbaiyan 
Its requires many configuration on the bigip like VIP,pool and cookie needs to be created.
1. When the user access the url through a client like browser, it connects to DNS to get the ip of the domain name which should be pointing to a VIP in the bigi
p, then the request comes to bigip based on the ip resolved by the DNS.
2. The VIP forwards that request to one of the available apache server in the pool based on the load balancing algorithm's ( round robin,least connection method ..etc).
3. Then Apache forward that request on the available nodes in the weblogic cluster, then the response will be send back to the apache.
4. Apache will send the response back to BIGIP, then BIGIP will send the response to the client.



1 comment:



  1. Informative post. But OTD is a software loadbalancer, not a h/w.

    ReplyDelete