SYSTEM GENERATION

An operational system is a combination of the z/TPF system, application programs, and people. People assign purpose to the system and use the system. The making of an operational system depends on three interrelated concepts:

- System definition: The necessary application and z/TPF system knowledge required to select the hardware configuration and related values used by thez/TPF system software.

- System initialization: The process of creating the z/TPF system tables and configuration-dependent system software.

- System restart and switchover: The procedures used by the z/TPF system software to ready the configuration for online use.

The first two items are sometimes collectively called system generation; also installing and implementing. System definition is sometimes called design. System restart is the component that uses the results of a system generation to place the system in a condition to process real-time input. The initial startup is a special case of restart and for this reason system restart is sometimes called initial program load, or IPL. System restart uses values found in tables set up during system generation and changed during the online execution of the system. A switchover implies shifting the processing load to a different central processing complex (CPC), and requires some additional procedures on the part of a system operator. A restart or switchover may be necessary either for a detected hardware failure, detected software failure, or operator option. In any event, system definition (design), initialization, restart, and switchover are related to error recovery. This provides the necessary background to use this information, which is the principal reference to be used to install the z/TPF system.

Performing a system generation requires a knowledge of the z/TPF system structure, system tables, and system conventions, a knowledge of the applications that will be programmed to run under the system, and a user's knowledge of z/OS. Knowledge of the z/TPF system, Linux, and the application are required to make intelligent decisions to accomplish the system definition of a unique z/TPF system environment. The use of z/OS and Linux is necessary because many programs used to perform system generation run under control of z/OS or Linux. Although this information does not rely on much z/OS or Linux knowledge, when the moment arrives to use the implementation information, the necessary z/OS and Linux knowledge must be acquired. You are assumed to have some knowledge of the S/370 assembly program as well as jargon associated with the z/OS and Linux operating systems. Some knowledge of C language is also helpful, because some of the programs that are used to generate the system are written in C.

System definition is a nontrivial endeavor because both z/TPF system and application knowledge is necessary. An understanding of basic data organization of the system is necessary to allocate the appropriate area of online file storage for all the programs. All z/TPF system programs and application programs that are used in the online environment are placed in file storage and reside in core memory.

An appreciation of defining a system is necessary to relate a unique z/TPF system to its environment. Clearly, no amount of information can anticipate all the variants of applications to be run under the system. However, the interrelationship of the applications, the z/TPF control program, and the corresponding data structures is of paramount importance. The initial system definition must be approached with the best available information. Usually, even this is not sufficient for an accurate definition. Through knowledge, analysis, and an understanding of system guidelines, a workable operational system can be defined and initialized. A system definition is not as insurmountable as it may first seem, because the z/TPF system includes components that allow the system programmer to change the system tables and programs, the application programs, and the physical components. In this sense, the operational system can be adapted to its environment.

The initial system generation, which requires analysis techniques and knowledge of the z/TPF system, represents the beginning of a continuous process required throughout the life of an operational system.

The terms system initialization and initialization process are used to emphasize the human activity and offline procedures required to produce programs and data to be placed on online system resident storage. The term offline procedures refers to the execution of programs, written to run under z/OS or Linux, that assist in the production of the online programs and data. (The term procedures is generally used to mean a program or collection of programs.) The goal of system initialization is to place the various levels of z/TPF system storage facilities into the condition where restart procedures may be invoked to allow online processing to begin. The restart procedures are involved with the programs that build system records used to control online processing. Many of these records are assigned values identified by the initialization process. Some important distinctions of terminology follow:

- A program in the restart procedures, called the initializer program (thez/TPF system identification is CCCTIN), uses values assigned to system records used for main (core) storage management. The initializer program is not the same thing as the initialization process. The initializer program is a portion of the restart procedures.

- A system restart begins by pressing the initial program load (IPL) key. This invokes a standard S/390 IPL sequence, which in turn invokes the z/TPF system IPL program. The IPL program loads enough of the system to main storage to permit the initializer program to allocate system tables to main storage. In a loosely coupled complex, an IPL requires the coordination of each of the CPCs in the complex; if a CPC is a multiprocessor, the IPL is performed by only one of the CPUs.

An initial restart differs from an online restart, which is shown in Figure 1. An initial restart uses input data and restart programs created in the offline environment. An online restart uses some of the same data, but the data and restart programs are accessed from online files. The basic restart procedure of loading the z/TPF core resident programs and executing the initializer program is identical for both the initial restart or the online restart. The location of the restart programs and data identifies the difference. During an initial restart, additional programs to load programs and data to system storage must be invoked. Normally, much of the data put in place during an initial restart does not need to be reloaded during an online restart. (If a reload is necessary, an initial restart is required.) The basic restart programs (such as the IPL and initializer programs) and the structure of the data that these restart programs process are identical. The location of the data is different, as well as some of the data content.

Figure 1. Initial vs. online restart