Processes, Scheduling, and Performance
Processes are created with RUN (new process) or SPAWN (subprocess), and controlled with STOP, SUSPEND, RESUME, and SET PROCESS. Each process has a priority (0-15 base, with real-time priorities available to privileged users), a quantum, a working set, and quotas (e.g., BIOLM, DIOLM, ASTLM, FILLM, PGFLQUOTA). The scheduler is priority-based with round-robin among equal priorities; candidates must understand that a higher base priority preempts lower ones and that real-time priorities (16-63) bypass normal scheduling.
Performance analysis uses MONITOR (classes such as CPU, MEMORY, DISK, IO), SHOW PROCESS/CPU, and the process accounting file. Key concepts: working set size and the free-page list, page faults (hard vs. soft), and the balance set. Excessive paging indicates insufficient memory or an oversized working set; high Direct I/O rates suggest disk bottlenecks.
Traps: confusing base priority with current priority (which can be temporarily boosted by the scheduler); assuming SUSPEND frees resources (it does not - the process remains); and forgetting that quotas are per-process limits, not system-wide. Also know that an image can be installed with privileges (INSTALL) to run with elevated privilege without granting the user that privilege - a common performance and security question.
Rule of thumb: when a process is slow, check first whether it is waiting on a resource (SHOW PROCESS ... /CONTINUOUS) before blaming CPU priority; priority only matters when the CPU is contended.