Beluga: A CXL-Based Memory Architecture for Scalable and Efficient LLM KVCache Management

Summary: Beluga exploits CXL switches to expose a shared, large-scale memory pool with native load/store access for GPU/CPU KVCache, avoiding RDMA’s latency/protocol overhead. Beluga-KVCache uses this architecture to scale long-context LLM inference, cutting TTFT 89.6% and boosting vLLM throughput 7.35x. (summarized by gpt-5-mini on Apr 11 2026)

Paper ID: 7454
Venue: SIGMOD
Year: 2026
Pagerank: 5.1725247e-05
Overall Rank: 10,143 | 29.51%
DOI: 10.1145/3786627

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

Incoming Citations (Sorted by Pagerank)

Showing 0 of 0 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank

Outgoing Citations (Sorted by Pagerank)

Showing 17 of 17 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
1,266	PolarFS: An Ultra-low Latency and Failure Resilient Distributed File System for Shared Storage Cloud Database	2018	VLDB	0.00011494173
1,960	ReAcTable: Enhancing ReAct for Table Question Answering	2024	VLDB	9.4913606e-05
2,007	Efficient Distributed Memory Management with RDMA and Caching	2018	VLDB	9.3992618e-05
3,200	Rethinking Database High Availability with RDMA Networks	2019	VLDB	7.7172767e-05
3,362	OpenSearch-SQL: Enhancing Text-to-SQL with Dynamic Few-shot and Consistency Alignment	2025	SIGMOD	7.547454e-05
3,727	ScaleStore: A Fast and Cost-Efficient Storage Engine using DRAM, NVMe, and RDMA	2022	SIGMOD	7.2310205e-05
4,260	Design Guidelines for Correct, Efficient, and Scalable Synchronization using One-Sided RDMA	2023	SIGMOD	6.8645333e-05
5,518	Distributed GPU Joins on Fast RDMA-capable Networks	2023	SIGMOD	6.2532393e-05
6,291	DEX: Scalable Range Indexing on Disaggregated Memory	2024	VLDB	5.9874246e-05
6,397	InferDB: In-Database Machine Learning Inference Using Indexes	2024	VLDB	5.9545897e-05
6,600	Serving Deep Learning Models with Deduplication from Relational Databases	2022	VLDB	5.8861676e-05
7,190	SpareLLM: Automatically Selecting Task-Specific Minimum-Cost Large Language Models under Equivalence Constraint	2025	SIGMOD	5.7341493e-05
7,687	Rethinking Stateful Stream Processing with RDMA	2022	SIGMOD	5.6249337e-05
8,473	Zero-sided RDMA: Network-driven Data Shuffling for Disaggregated Heterogeneous Cloud DBMSs	2024	SIGMOD	5.4888649e-05
8,849	Unlocking the Potential of CXL for Disaggregated Memory in Cloud-Native Databases	2025	SIGMOD	5.4233138e-05
9,472	Adda: Towards Efficient in-Database Feature Generation via LLM-based Agents	2025	SIGMOD	5.3246578e-05
10,788	From Scale-Up to Scale-Out: PolarDB's Journey to Achieving 2 Billion tpmC	2025	VLDB	5.1725247e-05

Semantically Similar Papers

Overall Rank	Paper	Year	Venue	Pagerank
10,222	RetroInfer: A Vector Storage Engine for Scalable Long-Context LLM Inference	2026	VLDB	5.1725247e-05
10,066	DepCache: A KV Cache Management Framework for GraphRAG with Dependency Attention	2026	SIGMOD	5.1725247e-05
13,152	Database Perspective on LLM Inference Systems	2025	VLDB	-
8,497	CXL Memory Performance for In-Memory Data Processing	2025	VLDB	5.4860111e-05
13,101	OrbitFlow: SLO-Aware Long-Context LLM Serving with Fine-Grained KV Cache Reconfiguration	2026	VLDB	-
4,681	Cache-Craft: Managing Chunk-Caches for Efficient Retrieval-Augmented Generation	2025	SIGMOD	6.6286556e-05
10,020	HotPrefix: Hotness-Aware KV Cache Scheduling for Efficient Prefix Sharing in LLM Inference Systems	2026	SIGMOD	5.1725247e-05
9,777	MEMO: Fine-grained Tensor Management For Ultra-long Context LLM Training	2025	SIGMOD	5.2743647e-05
8,849	Unlocking the Potential of CXL for Disaggregated Memory in Cloud-Native Databases	2025	SIGMOD	5.4233138e-05
6,185	PQCache: Product Quantization-based KVCache for Long Context LLM Inference	2025	SIGMOD	6.0213899e-05