Efficient memory management for large language model serving with pagedattention
High throughput serving of large language models (LLMs) requires batching sufficiently
many requests at a time. However, existing systems struggle because the key-value cache …
many requests at a time. However, existing systems struggle because the key-value cache …
ZenFS+: Nurturing performance and isolation to ZenFS
This paper proposes ZenFS+, a new storage backend of RocksDB for small-zone ZNS SSD.
RocksDB has complicated internal operations such as flush and compaction. Flush and …
RocksDB has complicated internal operations such as flush and compaction. Flush and …
Prism: Optimizing key-value store for modern heterogeneous storage devices
As data generation has been on an upward trend, storing vast volumes of data cost-
effectively as well as efficiently accessing them is paramount. At the same time, today's …
effectively as well as efficiently accessing them is paramount. At the same time, today's …
Revisiting Secondary Indexing in {LSM-based} Storage Systems with Persistent Memory
LSM-based storage systems are widely used for superior write performance on block
devices. However, they currently fail to efficiently support secondary indexing, since a …
devices. However, they currently fail to efficiently support secondary indexing, since a …
Replicating Persistent Memory {Key-Value} Stores with Efficient {RDMA} Abstraction
Combining persistent memory (PM) with RDMA is a promising approach to performant
replicated distributed key-value stores (KVSs). However, existing replication approaches do …
replicated distributed key-value stores (KVSs). However, existing replication approaches do …
MoltDB: Accelerating Blockchain via Ancient State Segregation
J Liang, W Chen, Z Hong, H Zhu… - IEEE Transactions on …, 2024 - ieeexplore.ieee.org
Blockchain store states in Log-Structured Merge (LSM) tree-based database. Due to
blockchain traceability, the growing ancient states are inevitably stored in the databases …
blockchain traceability, the growing ancient states are inevitably stored in the databases …
Perseid: A Secondary Indexing Mechanism for LSM-Based Storage Systems
LSM-based storage systems are widely used for superior write performance on block
devices. However, they currently fail to efficiently support secondary indexing, since a …
devices. However, they currently fail to efficiently support secondary indexing, since a …
PetPS: Supporting huge embedding models with persistent memory
Embedding models are effective for learning high-dimensional sparse data. Traditionally,
they are deployed in DRAM parameter servers (PS) for online inference access. However …
they are deployed in DRAM parameter servers (PS) for online inference access. However …
TrieKV: A High-Performance Key-Value Store Design with Memory as Its First-Class Citizen
H Sun, D Kong, S Jiang, Y Yue… - IEEE Transactions on …, 2024 - ieeexplore.ieee.org
Key-value (KV) stores based on log-structured merge tree (LSM-tree) have been extensively
studied and deployed in major information technology infrastructures. Because this type of …
studied and deployed in major information technology infrastructures. Because this type of …
Optimizing File Systems on Heterogeneous Memory by Integrating {DRAM} Cache with Virtual Memory Management
Y Liu, Y Ren, M Liu, H Li, H Guo, X Miao, X Hu… - … USENIX Conference on …, 2024 - usenix.org
This paper revisits the usage of DRAM cache in DRAM-PM heterogeneous memory file
systems. With a comprehensive analysis of existing file systems with cache-based and DAX …
systems. With a comprehensive analysis of existing file systems with cache-based and DAX …