Clustering clarification needed please ....

Question

i am working on webcenter content clustering .
i have to hosts (host1 and host2 ) ,
-installed weblogic and UCM on both hosts with the same directory structure .
-create domain in host 1 ,add a cluster ,complete everything and start ucm on host 1 ,then i used pack and unpack utility to copy domain config from host1 to host 2 .
-start ucm on both hosts ,add document using ucm on host 1 and i can search for it from host 2 .
now ,this is called active -active clustering right ? ??
my managers needs my cluster to work as follow :
if any user where working on ucm (host1) and something goes wrong with ucm on host1 ,clients should be able to continue their work using ucm on host 2 .without reloading the page .
-is this possible with active active clustering ? which product will control this ?
-another question please ,the cluster ware is on host 1 ,so if anything goes wrong with host one ,then whole cluster will be down right ???? so what is the best scenario to do this ?
-and http server ,what it can help in clustering ?
some clustering expert explain these points to me please ,i am reading but not able to connect things together well !!!
thanks in advance .....

Answer 1

First, I will assume that your architecture is as follows:
Host 1 - Running two weblogic server instances (AdminServer and UCM_Server1)
Host 2 - Running one weblogic server instance (UCM_Server2)
Now, when you open the WebLogic console, you should see that both UCM_Server1 and UCM_Server2 are part of the same cluster.
Since you will have your UCM application deployed to the cluster (UCM_Server1 and UCM_Server2), it should be available and running on both the managed servers.
And this is exactly what clustering means, the user will be able to continue their work on server2 when failed over from server1
In weblogic we only have active-active clustering and your requirement is the typical usecase of a cluster
While the continuation of the work is controlled by the replication activity between the managed servers (UCM_Server1 and UCM_Server2) which occurs in the background, the actual failover needs to be performed by the WebServer.
WebServer significance:
Let's say we did not have a WebServer, in the browser when you provide a URL to access the application, it will look as below:
http://UCM_Server1sIPorDNS:Port/ucm
Now, what will happen if host1 is down or UCM_Server1 is down, your browser will not know where to go to and user will also realize that a backend server has failed, which we do not want to happen.
So, the ideal architecture is
Client (Browser) --> WebServer (OHS) --> WebLogicServer (UCM_Server1 OR UCM_Server2)
In this case, the URL in the browser will be the WebServer URL.
http://WebServerIPorDNS:Port/ucm
Now, even when UCM_Server1 is down, client will never know as the webserver will automatically failover the request to UCM_Server2.
Since, UCM_Server1 and UCM_server2 are in a cluster, the session information is available in UCM_Server2 for the request/transaction to continue from where it stopped processing on UCM_Server1
ONE MORE CLARIFICATION:
If you understood the above clarification, you will realize that cluster is virtual (not a physical process).
So, even if host1 goes down, there is no issue as UCM_Server2 (part of the cluster) will continue to serve the request of the appication.
WebServer will migrate/failover all the user requests that are being currently served by UCM_Server1
NOTE: It is recommended to run your webserver (if you are using a single webserver only) on a host other than the hosts running the UCM servers.
Hope this helps to answer your queries.
Please let me know if you have any further questions