CA Gen: two bases, seats and host capacity.

CA Gen is a model driven application development environment from Broadcom (CA), a CASE tool with roots in the earlier Texas Instruments and Sterling lineage. Developers define an application at a high level in a model held in an encyclopedia, and CA Gen generates the running code, COBOL, C, or Java, from that model. Large enterprises built core systems in it, and because those applications are generated rather than hand written, the model and the tool stay in the estate for the life of the application. That longevity is the backdrop to every CA Gen renewal.

CA Gen commonly carries two licensing bases. The development side, the modeling toolset and encyclopedia, is typically licensed by seat or named developer, scaling with the size of the team building and maintaining models. The generated runtime on the mainframe is licensed on host capacity in MIPS or MSU, scaling with the machine the applications run on. Either or both can fall under Broadcom Mainframe Consumption Licensing as part of a portfolio subscription. The precise split, and the capacity basis, is defined in your agreement, and the two streams should be modeled separately.

The runtime base hinges on the MIPS to MSU question; the development base hinges on seat counts. See the MIPS to MSU conversion question.

The first driver is developer seat count, which drifts as teams reorganize, contractors come and go, and access is granted but rarely revoked, so paid seats can outrun active users. The second is host capacity, since the runtime charge tracks the machine the generated applications run on and a hardware upgrade lifts it independent of any change in development activity. The third is the bundle: CA Gen frequently renews alongside other CA development and DevOps tools, where portfolio uplifts apply to everything at once and unused tools quietly carry forward.

Two bases mean two places exposure builds. Common traps we see at pattern level:

The levers split across the two bases plus the bundle. The five that pay:

There is no like for like swap for CA Gen while the generated applications are in production, because the applications are built from the model. The credible long term path is modernization: replatforming the generated COBOL or Java as maintainable standalone source, or rebuilding the applications on a current stack, both multi year programs with real risk. As a renewal lever, a modernization assessment earns its keep by being credible and documented, not by being executed mid term. Treat any pitch that frames a quick CA Gen exit with caution; the reengineering usually outweighs the near term saving.