The machine and the system
Mainframe. A large server built for high-volume transaction and batch work with very high availability. In this glossary it means IBM Z hardware running z/OS.
z/OS. IBM's main operating system for IBM Z mainframes.
LPAR. A logical partition: one slice of a mainframe running its own copy of an operating system. A single machine usually runs several LPARs, for example production, test and development, and a provider's estate may run hundreds across many clients.
Sysplex. A group of z/OS systems that work together and share resources. A Parallel Sysplex adds a coupling facility so the systems can share data.
Started task. A program that runs in its own address space, started with the START command from a procedure in a procedure library rather than submitted as a batch job. Subsystems such as JES2 and Db2 run as started tasks, and so do most CICS regions.
Messages and the console
Console. Where z/OS writes its messages and where operators type commands. Messages carry an identifier such as IEF238D that says which component issued them.
WTO. Write to operator: a message to the console that needs no answer.
WTOR. Write to operator with reply: a message that waits for an operator's answer, shown with a reply number, and answered with the REPLY command (R nn,text). Whatever issued it is held until the answer comes. See WTOR automation on z/OS.
SYSLOG. The system log: the record of console messages and commands.
SDSF. A panel-based tool for looking at jobs, output, the spool and the console on z/OS.
Jobs and batch
JCL. Job Control Language: the statements that describe a batch job, its steps, its programs and its data sets.
Job and step. A batch job is a unit of work made of one or more steps; each step runs one program.
JES2 and JES3. The job entry subsystems that receive jobs, schedule them onto initiators and manage their output.
Spool. The space where JES keeps job input and output. A spool filling up is an operations emergency.
Initiator. An address space that runs batch jobs, one at a time, for the job classes it is set to. When all initiators for a class are busy, jobs in that class wait.
Abend. An abnormal end of a program or step. System abend codes start with S (for example S0C7, a data exception, or S322, a CPU time limit), and user abend codes set by the application start with U.
Scheduler. Software that runs batch jobs in the right order at the right time, from triggers, requirements and calendars. Broadcom CA 7, BMC Control-M and IBM Z Workload Scheduler are common on z/OS.
Batch and batch SLA. A batch is a named chain of jobs that must finish together, such as a night's settlement. Its SLA is the time by which it must be done. See predicting batch SLA breaches.
Critical path. The longest chain of dependent jobs from now to the end of a batch. It decides when the batch ends.
Subsystems
CICS. IBM's transaction server for online applications. It runs in regions, which are address spaces.
IMS. IBM's hierarchical database and transaction manager.
Db2. IBM's relational database for z/OS.
Adabas. Software AG's database, used on z/OS by many long-running applications.
Measurement and money
SMF. System Management Facilities: z/OS's records of what happened, numbered by type. Type 30 covers jobs and steps; types 70 to 79 come from RMF performance monitoring.
RMF. Resource Measurement Facility: IBM's performance monitor for z/OS.
MSU and MIPS. Measures of mainframe capacity. MSU (millions of service units per hour) is what much mainframe software is priced on; MIPS is the older, informal measure.
Operations practice
Runbook. The written procedure for handling a situation: what to check, what to do, who to tell.
Four-eyes. A rule that a second person approves an action before it is sent.
Managed service provider. A company that runs other companies' mainframes under contract, usually with SLAs and credits for missed ones.
Meridian puts the console, the scheduler, the tickets and the batch SLA line for every LPAR on one screen. See what it does.