CONCEPT Cited by 1 source
Write endurance (NAND)¶
Definition¶
Write endurance of NAND flash is the number of Program/Erase (P/E) cycles a cell can survive before its bit-error-rate (BER) crosses the ECC-correctable threshold. Each write cycle incrementally degrades the tunnel oxide in the floating-gate transistor; beyond the endurance ceiling the cell becomes unreliable.
Endurance decreases as bits-per-cell increase, because more precise voltage discrimination is required. Approximate industry ceilings:
| Media | P/E cycles |
|---|---|
| SLC | ~100,000 |
| MLC (2-bit) | ~10,000 |
| TLC (3-bit) | ~3,000-5,000 |
| QLC (4-bit) | ~500-1,500 |
Why this was the QLC blocker¶
Meta's 2025-03-04 framing:
"QLC flash as a technology has been around since 2009. Adoption has been slow because it has historically operated at lower drive capacity points — less than 32TB. As well, high cost and limited write endurance didn't make it an attractive alternative to TLC in the datacenter."
At 2010s-era drive capacities, QLC's low P/E ceiling meant a mixed workload (say, the standard 3 DWPD / 5-year datacenter spec) would burn through the cells. Shipping QLC to production required the durability gap to close.
Two levers closed the gap¶
Lever 1: density increases shifted the math. A modern QLC drive has 10-30× more capacity than its 2010s predecessor. The absolute write volume per day stays the same (workloads don't write faster because the drive got larger), but spread over far more cells → per-cell P/E pressure drops.
Lever 2: workload matching. Meta's 2025 stance:
"QLC flash introduced as a tier above HDDs can meet write performance requirements with sufficient headroom in endurance specifications. The workloads being targeted are read-bandwidth-intensive with infrequent as well as comparatively low write bandwidth requirements."
Deploy QLC under read-dominant, low-write-BW workloads whose write volume is well below the cell's P/E budget. Durability is met by definition — the workload doesn't push enough writes to consume the endurance.
The endurance-power composition¶
The same property that makes QLC a bad fit for high-write workloads makes it a good fit for power-efficiency:
"Since the bulk of power consumption in any NAND flash media comes from writes, we expect our workloads to consume lower power with QLC SSDs."
Write-light workloads → fewer P/E cycles → lower write power → less energy spent per byte served. The endurance story and the power-efficiency story are two sides of the same coin.
Wear-leveling¶
Endurance is managed at drive / FTL granularity via wear-leveling — the controller distributes logical writes across all physical cells so no cell is exhausted before others. Host-side FTLs (see DirectFlash) give software direct control of wear-leveling policy.
Seen in¶
- sources/2025-03-04-meta-a-case-for-qlc-ssds-in-the-data-center — canonical framing of QLC's endurance-historically-blocking / workload-match-now-sufficient story.
Related¶
- systems/qlc-flash — the media this concept most constrains.
- systems/tlc-flash — higher-endurance sibling.
- concepts/bandwidth-per-terabyte — paired with endurance in matching workloads to media.
- concepts/storage-media-tiering — endurance is part of the tier-ordering rationale.
- patterns/middle-tier-storage-media.
- companies/meta.