发表机构
Harvard University; Brown University; Tufts University（哈佛大学; 布朗大学; 塔夫茨大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对ZNS SSD上AI检查点写密集导致的空间回收瓶颈，提出基于保留策略的回收时间感知分配器RetainZ，减少空间浪费与延迟，降低写放大。
AI 中文摘要
固态硬盘（SSD）正成为性能关键型存储的主导介质。分区命名空间（ZNS）SSD因顺序写入和显式分区重置可降低地址映射、过度配置和内部垃圾回收成本而日益受到青睐。然而，回收空间需要重置整个分区，因此删除一个文件并不会释放其空间，而同一分区内的其他文件必须保留。这种先擦除后写入的约束成为写密集型工作负载的瓶颈，尤其是在模型训练期间的AI检查点场景中，频繁保存模型和优化器状态会与长期保留的检查点竞争空间。我们观察到，保留策略本质上编码了数据何时可被淘汰的信息，从而无需学习生命周期预测器。我们提出RetainZ，一种将这些策略转化为显式回收周期的存储后端。其分配器隔离长期保留的数据，并将回收周期相近的对象分组，在延迟空间回收与分区末尾未使用空间之间取得平衡。在FEMU上使用多租户流和直至Pythia-1B的检查点清单进行评估，与成熟的生命周期感知ZNS后端ZenFS的分配器相比，在84%目标负载下，RetainZ将已删除数据和未使用分区尾部占用的空间减少41.7%，每个输入检查点的平均p99延迟降低47.5%，并将主机写放大从1.165降至1.003。
英文摘要
Solid State Drives (SSDs) are becoming the dominant medium for performance-critical storage. Zoned Namespace (ZNS) SSDs are getting more and more attractive because sequential writes and explicit zone resets reduce address-mapping, over-provisioning, and internal garbage-collection costs. However, reclaiming space requires resetting an entire zone, so deleting one file does not free its space while other files there must be kept. This erase-before-write constraint becomes a bottleneck for write-intensive workloads, particularly AI checkpointing during model training, where frequent saves of model and optimizer state compete for space with long-retained checkpoints. We observe that retention policies inherently encode when data can be retired, avoiding the need for a learned lifetime predictor. We present RetainZ, a storage backend that translates these policies into explicit reclaim epochs. Its allocator isolates long-retained data and groups objects with nearby reclaim epochs, balancing delayed space reclamation against unused space at zone ends. Evaluated on FEMU with multi-tenant streams and checkpoint manifests up to Pythia-1B, RetainZ reduces space occupied by deleted data and unused zone tails by 41.7% and mean per-input checkpoint p99 latency by 47.5% at 84% target load compared with the allocator of ZenFS, an established lifetime-aware ZNS backend, and lowers host write amplification from 1.165 to 1.003.
Comments5 pages, 4 figures. Code and reproducibility artifacts: https://github.com/DonaldLucy/RetainZ
