warm · research · also Write-hotness Aware Retention Management

WARM

WARM is a flash-management research design that separates frequently overwritten data into a distinct block pool to reduce retention-refresh writes.

Two writes within a recent-history window can promote a page into the hot pool; aging hot pages are demoted. Sequential allocation permits block-granularity tracking. The controller adjusts the pool boundary using write activity. [1]

Only the hot pool initially uses shorter retention without refresh. Combining the design with adaptive refresh later extends that treatment to cold blocks. Migration and garbage collection still consume writes; eliminating refresh is not eliminating maintenance. [1]

The evaluation uses DiskSim and repeated traces, comprising 14 recorded workloads plus two synthetic benchmarks. Estimated lifetime includes host, reclamation and refresh writes. It is not a longitudinal fleet observation. [1]

Relations

related
NAND retention relaxation, Flash Correct-and-Refresh, Flash data retention, Flash translation layer (FTL)

Limitations

  • Write-hotness is a workload property, not a powered-off retention guarantee. Historical modeled lifetime gains do not establish endurance for current SSDs. [1, 2]

Citations

  1. Luo et al. WARM. MSST 2015, sections 4-6 and Tables 1-2. DOI: 10.1109/MSST.2015.7208284. (accessed 2026-11-18)
  2. Cai et al. Flash Correct-and-Refresh. ICCD 2012, section IV.D: power supply continuity. (accessed 2026-11-18)
Sources & caveats
  • Literature synthesis; no controller implementation, physical-device test or replication of the paper's lifetime estimates was performed.

JSON: /data/terms/warm.json

Published by Hesela team · Updated

Report a correction · Editorial standards · Reuse guidance

Cite this page
Hesela team. WARM. Hesela. 2026-11-18. https://hesela.dev/terms/warm/