NonStop Server Guide for BEA WebLogic Server 8.1
Using the WS Plug-in
HP NonStop Server Guide for BEA WebLogic Server 8.1—529891-001
8-2
Routing to Replicated Applications
Routing to Replicated Applications
The WS Plug-in supports pathname prefix mapping, which routes requests to the same
application deployed in multiple clusters. To use this feature, you must replace the first
component of the pathname specified by the client with a different prefix before
forwarding it to the server. The WS Plug-in uses the prefix to determine to which
cluster to route the request and then it replaces the prefix specified by the client with
the one required by the server. This means that you can route preferred clients to a
separate cluster of application servers (which may be running at a higher priority) in
order to provide them with superior service.
Load Balancing
When the WS Plug-in is operating in J2EE mode, you have the option of using a load-
balancing algorithm based on POST directives and GET directives that include a query
string (because these requests are likely to involve significant processing by the
application server). The WS Plug-in also keeps track of the number of new sessions
allocated to each application server. Then it uses these two pieces of information to
determine the capacity of each target application server process in the cluster and
routes requests accordingly. This feature can be activated using a run time
environmental parameter. The default load balancing algorithm is 'round-robin'.
Tracking Failed Target Application Server Processes
The WS Plug-in also uses the J2EE mode setting to activate the recording and retrying
of failed target application server processes. If J2EE mode is configured, the router for
iTP will record failed servers in the DBSTATUS table and update the DBFILES table to
indicate to other router for iTP processes that a target application server has failed.
This action causes each router for iTP process to read the list of failed target
application servers in the DBSTATUS table and update its own internal table.
Passthrough and Redirect Modes
The WS Plug-in supports two types of operation: passthrough and redirect.
•
In passthrough mode, all requests are sent from the client to the iTP Secure
WebServer, then to the WS Plug-in and the application server. To use this mode,
set the noredirect column in the DBACCESS table to a positive value.
•
In redirect mode, the client initially communicates with the iTP Secure WebServer
and the WS Plug-in. However, having selected a server process, the WS Plug-in
instructs the client to communicate directly with that specific application server. In
this case, the client sends all subsequent requests directly to the specific target
application server process in the application server cluster selected by the WS
Plug-in. To use this mode, set the noredirect column in the DBACCESS table to
a negative value.
HP recommends the use of passthrough mode. This ensures that static content is
processed by iTP Secure WebServer rather than the application server and is also