slowdown-fairness · research · also FLIN strict fairness

Slowdown fairness

Slowdown fairness compares the smallest and largest relative slowdowns among concurrently running I/O flows.

FLIN defines S_i as a flow's shared mean response time divided by its isolated mean, then F = min(S_i) / max(S_i). Compare equal-priority flows at matching timing boundaries. [1]

Equal ratios can describe equally poor service. The constructed pairs (2, 2) and (20, 20) both give F = 1, despite tenfold different slowdowns; (8, 2) gives F = 0.25. Keep absolute latency and the flow set alongside the fairness value. [2]

Relations

related

Inter-flow interference, Storage latency, Slowdown ratio, Queue depth

Citations

1. Tavakkol et al. FLIN: Enabling Fairness and Enhancing Performance in Modern NVMe Solid State Drives. ISCA 2018, sections 2.2-6. (accessed 2026-10-20)
2. Hesela: Sharing an SSD: who pays for the slowdown? (accessed 2026-10-20)

JSON: /data/terms/slowdown-fairness.json