Why does something as simple as searching for an application's IP environment information need to be so complex? The answer lies only partially in the dual nature of the z/OS (MVS and UNIX) operating system.

The other reason is that some organizations have a need to run more than one TCP/IP stack within a single LPAR. In such a context, there are some applications (servers) that should be associated with one TCP/IP stack and not the other TCP/IP stack. There are many different ways that this can be controlled, and some of it depends upon the nature of the application itself. But for some applications it is controlled by the TCPIPJOBNAME statement found in the resolver configuration file.

This implies that an organization would need to have more than one resolver configuration file: one for application A, which uses TCP/IP stack X and another resolver configuration file for application B, which uses stack Y.

The other reason for the complexity--and this applies to essentially any aspect of z/OS--is the configurability requirements. The z/OS operating system is designed for the largest organizations in the world. Such organizations are often also less flexible, more process-oriented, and security-aware environments. Consequently, z/OS must fit into the organization's requirements, and not the other way around.

What does this really mean? Well, if XYZ Corporation wants to use a certain naming convention for certain data sets because thousands of users security profiles are written with this in mind, then z/OS TCP/IP had better be able to conform.