waiting time
The total amount of time in which a process spends in a ready queue before it reaches the CPU.
waiting time: queue delay before CPU execution
In a multitasking operating system, waiting time is the interval between when a process enters the ready queue and when the CPU scheduler actually assigns it processor cycles. During this period, the process is prepared to run, its context is loaded in memory, and its instructions are available, but the CPU is busy executing other work. This is distinct from execution time, which is the time the process actually spends on the CPU.
Waiting time accumulates in systems where many processes compete for a single CPU or a limited number of cores. In batch processing, a job might wait hours in queue before the scheduler picks it up. In real-time systems, waiting time is a critical constraint: a safety-critical process must not wait so long that it misses its deadline. In interactive systems, excessive waiting time causes visible lag or unresponsiveness from the user's perspective.
Measurement and scheduling impact
Waiting time is measured in milliseconds, seconds, or CPU cycles, depending on the context. The average waiting time across all processes is a key performance metric for evaluating scheduling algorithms. Different schedulers produce different waiting times: First Come First Served (FCCS) produces long average waiting times when short jobs queue behind long ones, while algorithms like Shortest Job First (SJF) or Round Robin minimize waiting by prioritizing short work or rotating fairly among contenders.
Waiting time is not caused by I/O delays, page faults, or system calls (those are typically called block time or sleep time). It arises purely from scheduler arbitration. In multiprocessor systems, waiting time can be reduced by having idle cores available, or by load balancing across cores. In systems with thread priority levels, high-priority threads experience shorter waiting times than low-priority ones.
System designers monitor waiting time using kernel profilers and tracing tools. Excessive waiting time signals either CPU overload, poor scheduler tuning, or processes being assigned the wrong priority class. In hard real-time embedded systems, waiting time is predicted and bounded during design; violating that bound is a failure.