MLFQ

Multi-Level Feedback Queue (MLFQ) is a CPU scheduling strategy used by computer operating systems to balance responsiveness for interactive tasks with overall system throughput. It organizes the ready jobs into multiple queues with different priorities and dynamically adjusts a process’s position based on its observed behavior.

Processes begin at the highest-priority queue. The scheduler always selects a job from the highest nonempty queue. Each queue has a distinct time quantum; short quanta favor quick, interactive tasks, while longer quanta in lower queues accommodate CPU-bound work. If a process uses its entire quantum in a given level, it is demoted to a lower-priority queue. If it yields the processor before its quantum expires or waits for I/O, it may stay in the same level or be promoted, depending on the implementation.

To prevent starvation, aging increases the priority of waiting processes after some time, gradually moving them toward higher queues.

Variants differ in the number of levels, quantum lengths, and promotion/demotion rules. The core idea is to provide responsiveness to short tasks while allowing long-running tasks to progress, without requiring fixed priorities for every task.

Usage and evaluation: MLFQ has been widely discussed in operating system literature and is used in some time-sharing systems and embedded platforms. It can incur scheduling and maintenance overhead and requires careful tuning of levels and quantum values; if aging is neglected, low-priority tasks may suffer from starvation. Modern general-purpose OS often use alternative schedulers, but MLFQ remains a foundational concept in CPU scheduling research and teaching.