InfoCenter Home >
7: Multimachine management >
7.3: Redirecting servlets >
7.3.2: Servlet redirector

7.3.2: Servlet redirector

How the servlet redirector works

The servlet redirector is a dedicated, special-purpose application server that runs on the same machine as the Web server.  It forwards HTTP requests received by a Web server to one or more application servers.  The servlet redirector receives HTTP requests through a local OSE channel just like any other application server.   However, instead of processing the requests itself, it sends them to servlet engines running in remote application servers for processing. 

The WebSphere Application Server EJB facility is used to forward requests from the servlet redirector to a remote application server.  Each application server that accepts requests from the servlet redirector contains a special stateless session bean called the RemoteSRP bean.  The RemoteSRP bean exports a method that forwards the incoming HTTP request to the servlet engine for execution. 

The servlet redirector acts as an EJB client to the application server.  When it receives an HTTP request, it looks up the RemoteSRP bean in the target application server.   It then remotely invokes the forwarding method through Remote Method Invocation (RMI) over the Internet Inter-ORB Protocol (IIOP).  The HTTP request is then forwarded from the servlet redirector to the servlet engine in the target application server for processing.

Because the servlet redirector communicates with application servers through EJB client invocations, it can participate in workload management.  This allows it to distribute requests to cloned application servers and provides support for load balancing and failover.  See article 7.2, Managing workloads, for more information on the workload management service.

The servlet redirector supports WebSphere security for servlets and enterprise beans.   It also supports encrypted communications between the servlet redirector and the application server.  However, it does not support Network Address Translation (NAT) firewalls.

The servlet redirector can be configured in three different ways, depending on how administration is set up on the Web server machine:

These configurations are described in the rest of this article.

Thick servlet redirector

In this configuration, the machine where the servlet redirector runs is configured as a full WebSphere Application Server node with an administrative server and its associated processes.  The thick servlet redirector is administered like any other application server process.  It requires a database client to access the WebSphere administrative repository; a database user ID and password must be stored on the machine for use by the database processes.  The administrative server also requires at TCP connection to the remote database.  If the thick servlet redirector is being used with a firewall, a port must be opened for database traffic and the firewall must support IIOP communications. 

Article 7.1.3.6.4, Thick servlet redirector sample topology, gives a more detailed description of this configuration of the servlet redirector.

Thick servlet redirector with administrative agent

In this configuration, the administrative server on the machine where the servlet redirector runs is configured as an agent of an administrative server running on another machine.  The servlet redirector is administered through the remote administrative server.  The agent receives configuration and administration information from the remote administrative server.  Access to the WebSphere administrative repository is through the remote administrative server, not the agent.  This reduces the number of processes that run on the Web server machine.  It also eliminates the need to install a database client on the machine.  If a firewall is in use, running the servlet redirector with an administrative agent instead of a full administrative server eliminates the need to open up a port for database communications.

Article 7.1.3.6.5, Thick servlet redirector with administrative agent sample topology, gives a more detailed description of this configuration of the servlet redirector.

Thin servlet redirector

In this configuration, there is no administrative server or agent on machine where the servlet redirector runs.  Instead, scripts are used to configure the Web server plug in to communicate with the servlet redirector, start the servlet redirector, and stop the servlet redirector.  The thin servlet redirector is harder to administer than other servlet redirector configurations.  However, because no administrative server or agent processes are running on the Web server machine, more processing power is available to handle HTTP requests.

Article 7.1.3.6.6, Thin servlet redirector sample topology, gives a more detailed description of this configuration of the servlet redirector.

Go to previous article: OSE and remote OSE Go to next article: WebSphere Application Server Problem Determination

 

 
Go to previous article: OSE and remote OSE Go to next article: WebSphere Application Server Problem Determination