Modern heterogeneous storage no longer follows a simple fast/slow hierarchy. Byte-addressable NVM provides fine-grained, low-latency persistence, whereas NVMe SSDs deliver high bandwidth under deep queues and offer greater capacity. Most LSM-tree key-value stores, however, send every foreground write through a designated performance tier; SSDs receive data only later through flush, compaction, or migration. When the NVM path saturates, this fixed entry point strands SSD write bandwidth. We present DuetKV, a heterogeneous key-value store that treats NVM and SSD as peer foreground write engines and selects a destination for each request. Saturation-aware routing favors NVM while it has headroom and shifts writes to SSD under pressure. A sync-async interleaved pipeline overlaps synchronous NVM persistence with asynchronous SSD I/O. Size-adaptive cross-tier compaction reclaims NVM space with low write amplification, while transparent read mirrors serve hot SSD-resident data from NVM. On our NVM–SSD prototype, DuetKV achieves up to 3.4 × the throughput of the strongest baseline on write-heavy workloads, while remaining competitive on read-heavy workloads.
Rui-Song Zhou, Liang Bao, Hai-Jun Zhang et al.· 0 citations
Adopting replicated persistent key-value stores (RPKVSs) as metadata backends is becoming increasingly popular in modern large-scale storage systems. Conventional RPC-based KV replication is infamously subject to its replication overheads due to the inherent architectural mismatch between the application-level network stack and storage-prioritized operations. In this paper, we shift toward a new paradigm of storage-native block-based replication to fundamentally address the limitations of conventional replication. This shift is made possible by using widely deployed NVMe-oF technology to unify replication and storage efficiency under integrated block semantics. We present NoFDB, an NVMe-oF-enabled RPKVS design, that eliminates redundant index management and reduces replication latency, and introduces a set of system-level optimizations to address the challenges introduced by NVMe-oF and unlock its performance potential. Our evaluation shows that NoFDB improves random write throughput by up to 2.85 × and macrobenchmark throughput by up to 3.10 × over the best-performing baseline in each setting.
Liang Bao, Rui-Song Zhou, Hua Wang et al.· ACM Transactions on Architec...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
