How WAFL in ONTAP Maximizes Flash Capacity: Dedup, Compression, Compaction, and TSSE

What this brief covers

This explains how NetApp's WAFL (Write Anywhere File Layout) integrates deduplication, compression, compaction, and TSSE to improve usable capacity on flash systems, especially QLC-based AFF/ASA C-Series platforms.

The pipeline: one ordered process, not three separate features

WAFL does dedupe, then compression, then compaction as a single inline pipeline before data is written to media. The order is deliberate:

• Deduplication runs first on whole 4KB blocks. It finds identical blocks before any shrinking occurs, avoiding missed matches that would happen if you compared compressed fragments.
• Compression runs next and reduces block size where possible.
• Compaction runs last to collect leftover sub-4KB pieces and pack them into a single physical 4KB block.

Because the stages work together in sequence, their gains are additive in practice: dedupe finds identical 4KB data vectors, compression reduces each vector, and compaction eliminates padding waste created by fixed-block writes.

The 4KB problem and why compaction matters

WAFL writes in fixed 4KB blocks, which aligns with flash geometry but creates inefficiency when data compresses below 4KB or applications write many small files. Without compaction, each sub-4KB fragment still consumes a whole physical block, wasting space.

Inline compaction, introduced in ONTAP 9, packs multiple sub-4KB chunks into a single 4KB physical block before the write occurs. That can turn multiple tiny writes into a single physical block consumption, recovering capacity that dedupe and compression by themselves couldn't reclaim.

TSSE: moving heavier compression off the critical path

Temperature-Sensitive Storage Efficiency (TSSE), added in ONTAP 9.8, uses a background process to apply different compression strategies by data temperature:

• Hot data (frequently overwritten) is compressed inline in small 8KB groups for speed and low latency impact.
• Cold data (typically after ~two weeks of inactivity) is scanned in the background and recompressed into larger 32KB groups with a stronger algorithm, which yields higher compression ratios but is heavier on CPU.

This design avoids putting the heavier decompression cost on the inline write path and applies it only where it's beneficial for density. TSSE is enabled by default on thin AFF volumes and on the C-Series.

Where hardware and software meet

Newer AFF platforms (A70, A90, A1K) can offload 32KB compression to a dedicated hardware processor and run that stronger compression continuously rather than relying solely on background cycles. That changes the performance/density trade-off but still relies on the same pipeline and compaction concepts to maximize usable capacity on QLC and other flash media.

Practical takeaway for storage buyers and operators

If you're buying QLC-based all-flash capacity, look beyond raw terabytes and check how the platform treats inline efficiency and background recompression. A pipeline that runs dedupe first, then compression, then compaction — combined with TSSE or hardware offload for cold data compression — will extract more usable capacity for real workloads without imposing extra latency on hot writes.