bursttime

Burst time, also seen in concatenated form bursttime in some texts, refers to the amount of time a process or thread is allowed to run on the CPU before it must wait for I/O or yield control. A process typically alternates between CPU bursts and I/O bursts; the length of each CPU burst is called its burst time. Burst times vary widely among processes and can be influenced by workload, system resources, and program design.

In scheduling, burst time is a key parameter. Algorithms such as shortest-job-first (SJF) aim to minimize waiting time by prioritizing processes with shorter burst times. Round-robin and other time-sharing schemes use fixed time slices, but the scheduler still benefits from knowledge or estimates of burst lengths to improve responsiveness and throughput. Because burst times are often unknown at arrival, systems rely on estimates derived from past CPU usage.

Estimation and measurement: The OS may estimate a process's next CPU burst length, often by using the length of the previous bursts and an exponential averaging formula. Incorrect estimates can lead to longer waiting times or unfairness. Burst time distributions can be unimodal or heavy-tailed, and workloads where long bursts are common can degrade performance for short-burst processes if not managed carefully.

In practice, burst time is an abstract metric used by schedulers to optimize CPU utilization and responsiveness. It is sometimes contrasted with I/O burst time, the time spent performing input/output operations, which typically requires the CPU less frequently.