Characterizing and Emulating FDP SSDs with WARP

Flexible Data Placement (FDP) promises to reduce write amplification by steering writes across reclaim unit handles (RUHs), yet outcomes vary widely across devices. This paper presents WARP, the first open emulator and comprehensive study of FDP SSDs. Our cross-device, cross-workload characterization shows that FDP sustains near-1 WAF when RUH isolation aligns with object lifetimes, but fails under misclassification, RUH interference, or adversarial invalidations. WARP reproduces hardware WAF trends while exposing per-RUH dynamics and configurable policies hidden in real devices. With WARP, we explore the firmware design space for FDP and demonstrate policies that reduce WAF beyond current hardware. By combining empirical characterization with a transparent emulator, this work advances FDP research from anecdotal reports to principled understanding and provides a platform for future FDP-aware system design.

Characterization. We provide the first cross-device, cross-workload study of two commercial FDP SSDs, revealing both their strengths and fragilities. FDP consistently lowers WAF in cache-like workloads, but it breaks under co-located traffic and adversarial invalidations. Our results show sharp vendor-dependent variability and uncover two previously undocumented phenomena: Noisy RUH, where invalidations in one handle amplify writes in others, and Save Sequential, where devices prematurely reclaim long sequential streams. Together, these findings show that FDP's promise of near-1 WAF is not guaranteed; it is workload-and configuration-dependent.

Emulation. To explain these effects, we build and validate WARP 1 , the first open FDP emulator. WARP faithfully reproduces real-device WAF trends while exposing internal dynamics that hardware conceals, such as per-RUH amplification, GC victim choices, and resource sharing between RUHs. Beyond validation, WARP turns opaque firmware defaults into tunable research knobs: II vs. PI isolation, RU size, OP ratio, and GC strategies. With this visibility, we systematically explore FDP's design space.

Our exploration yields new design-level understanding. We show that PI only outperforms II above device-dependent OP thresholds, while II is more resilient under limited slack ( §5). Such insights lead to further research questions. By releasing WARP as an open platform, we enable reproducible and full-stack FDP research spanning firmware, OS, and applications.

By combining broad characterization with a validated emulator, this work moves the community from anecdotal evidence toward a principled understanding of FDP. Our results show that FDP is flexible but not foolproof : it delivers when workloads align with its placement model, but can fail under others. WARP bridges this gap by providing both the empirical evidence and the mechanistic insights necessary to design FDP-aware systems and controllers.

Contributions. We make the following contributions:

• We conduct the first systematic study of commercial FDP SSDs across synthetic, trace-driven, and file-system workloads, revealing when FDP sustains near-1 WAF and when it collapses.

• We identify two previously unreported behaviors that explain how RUH interference and premature reclamation 1 WARP stands for Write Amplification Research Platform erode FDP's benefits.

• We design and validate WARP, the first open FDP emulator that faithfully reproduces hardware trends while exposing per-RUH amplification, GC victim choices, and tunable geometry.

• Using WARP, we explore II vs. PI, OP ratios, and RU sizing, showing that PI outperforms II only above device-dependent OP thresholds, while II is more robust under limited slack. We propose firmware strategies that reduce WAF beyond current hardware.

• We have upstreamed WARP to FEMU at https://gi thub.com/MoatLab/FEMU.