Definition
Plain language
A classic operating-system technique that gives short, fast jobs priority and lets long jobs run in the background.
As stated in the literature
A scheduling discipline using multiple priority queues with promotion and demotion rules, balancing responsiveness for short tasks against fairness for long ones.
Also called: MLFQ
Why it matters: It's how operating systems keep interactive workloads snappy without starving long-running jobs, a balance that pure FIFO can't achieve.
For example, your terminal commands return instantly while a long compile runs in the background because the scheduler kept demoting the compile.
Heard on the show
“The internal scheduler replaces first-come-first-served — which is what most LLM servers do by default — with a multi-level feedback queue.”Episode 016 — Why Your Coding Agent Stalls While the GPU Runs Hot