Learning DCOM

Endpoint Configuration

In traditional client/server systems that use TCP/IP for client/server communications, the server listens for client requests on a specific endpoint, which is more often referred to as a port. A client has to know the server IP address and the server port for successful communications with the server.

Unlike traditional client/server systems, COM dynamically activates components. Upon activation, COM dynamically assigns the launched component a connection endpoint, so a given component can use a different endpoint every time it is activated. For example, a connection endpoint for a given component may be 4000 at one time but be 5000 at another time.

Because component endpoints are dynamically assigned, it is painful to configure a component object for use through a firewall. You know that port 135 must be opened, because the Distributed COM infrastructure uses this port (See Figure 5-10 in Chapter 5). One port is understandable, but imagine going through the correct firewall/network/security people to open up a wide range of ports for a particular IP address. To ameliorate this problem, Windows NT 4.0 Service Pack 4 allows you to statically configure an endpoint for each AppID. Once configured, the component associated with this AppID will be activated using the specified endpoint every time.

The dcomcnfg.exe utility that comes with Windows NT 4.0 Service Pack 4 has a newly added property page called Endpoints, as shown in Figure C-1.

Figure C-1. The “Endpoints” ...