# TokenLake: A Unified Segment-level Prefix Cache Pool for Fine-grained Elastic Long-Context LLM Serving

arXiv: https://arxiv.org/abs/2508.17219 · HTML: https://mlsanity.com/paper/2508.17219/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 39 of its 61 references that are in the map.
- First posted on arXiv 24 Aug 2025.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

## Builds on (most cited first, 30 of 39)

- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Ray: A Distributed Framework for Emerging AI Applications](https://mlsanity.com/paper/1712.05889/) (2017)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [Ring Attention with Blockwise Transformers for Near-Infinite Context](https://mlsanity.com/paper/2310.01889/) (2023)
- [FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving](https://mlsanity.com/paper/2501.01005/) (2025)
- [World Model on Million-Length Video And Language With Blockwise RingAttention](https://mlsanity.com/paper/2402.08268/) (2024)
- [Cost-Efficient Large Language Model Serving for Multi-turn Conversations with CachedAttention](https://mlsanity.com/paper/2403.19708/) (2024)
- [FastMoE: A Fast Mixture-of-Expert Training System](https://mlsanity.com/paper/2103.13262/) (2021)
- [LoongServe: Efficiently Serving Long-Context Large Language Models with Elastic Sequence Parallelism](https://mlsanity.com/paper/2404.09526/) (2024)
- [Accelerating Distributed MoE Training and Inference with Lina](https://mlsanity.com/paper/2210.17223/) (2022)
- [DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference](https://mlsanity.com/paper/2401.08671/) (2024)
- [ChunkAttention: Efficient Self-Attention with Prefix-Aware KV Cache and Two-Phase Partition](https://mlsanity.com/paper/2402.15220/) (2024)
- [Infinite-LLM: Efficient LLM Service for Long Context with DistAttention and Distributed KVCache](https://mlsanity.com/paper/2401.02669/) (2024)
- [Preble: Efficient Distributed Prompt Scheduling for LLM Serving](https://mlsanity.com/paper/2407.00023/) (2024)
- [Stateful Large Language Model Serving with Pensieve](https://mlsanity.com/paper/2312.05516/) (2023)
- [SCBench: A KV Cache-Centric Analysis of Long-Context Methods](https://mlsanity.com/paper/2412.10319/) (2024)
- [KVFlow: Efficient Prefix Caching for Accelerating LLM-Based Multi-Agent Workflows](https://mlsanity.com/paper/2507.07400/) (2025)
- [Striped Attention: Faster Ring Attention for Causal Transformers](https://mlsanity.com/paper/2311.09431/) (2023)
- [FastDecode: High-Throughput GPU-Efficient LLM Serving using Heterogeneous Pipelines](https://mlsanity.com/paper/2403.11421/) (2024)
- [POD-Attention: Unlocking Full Prefill-Decode Overlap for Faster LLM Inference](https://mlsanity.com/paper/2410.18038/) (2024)
- [KVCache Cache in the Wild: Characterizing and Optimizing KVCache Cache at a Large Cloud Provider](https://mlsanity.com/paper/2506.02634/) (2025)
- [Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures](https://mlsanity.com/paper/2505.09343/) (2025)
- [NEO: Saving GPU Memory Crisis with CPU Offloading for Online LLM Inference](https://mlsanity.com/paper/2411.01142/) (2024)
- [Context Parallelism for Scalable Million-Token Inference](https://mlsanity.com/paper/2411.01783/) (2024)
- [Serving Large Language Models on Huawei CloudMatrix384](https://mlsanity.com/paper/2506.12708/) (2025)
- [Prefill-Decode Aggregation or Disaggregation? Unifying Both for Goodput-Optimized LLM Serving](https://mlsanity.com/paper/2508.01989/) (2025)
- [BlendServe: Optimizing Offline Inference for Auto-regressive Large Models with Resource-aware Batching](https://mlsanity.com/paper/2411.16102/) (2024)

## Built on by (most cited first, 5 of 5)

- [DualPath: Breaking the Storage Bandwidth Bottleneck in Agentic LLM Inference](https://mlsanity.com/paper/2602.21548/) (2026)
- [CONCUR: High-Throughput Agentic Batch Inference of LLM via Congestion-Based Concurrency Control](https://mlsanity.com/paper/2601.22705/) (2026)
- [WarmServe: Enabling One-for-Many GPU Prewarming for Multi-LLM Serving](https://mlsanity.com/paper/2512.09472/) (2025)
- [RepetitionCurse: Measuring and Understanding Router Imbalance in Mixture-of-Experts LLMs under DoS Stress](https://mlsanity.com/paper/2512.23995/) (2025)
- [NanoCP: Request-Level Dynamic Context Parallelism for Data-Expert Parallel Decoding](https://mlsanity.com/paper/2605.21100/) (2026)

## Abstract (arXiv)

Prefix caching is crucial to accelerate multi-turn interactions and requests with shared prefixes. At the cluster level, existing prefix caching systems are tightly coupled with request scheduling to optimize cache efficiency and computation performance together, leading to load imbalance, data redundancy, and memory fragmentation of caching systems across instances. To address these issues, memory pooling is promising to shield the scheduler from the underlying cache management so that it can focus on the computation optimization. However, because existing prefix caching systems only transfer increasingly longer prefix caches between instances, they cannot achieve low-latency memory pooling. To address these problems, we propose a unified segment-level prefix cache pool, TokenLake. It uses a declarative cache interface to expose requests' query tensors, prefix caches, and cache-aware operations to TokenLake for efficient pooling. Powered by this abstraction, TokenLake can manage prefix cache at the segment level with a heavy-hitter-aware load balancing algorithm to achieve better cache load balance, deduplication, and defragmentation. TokenLake also transparently minimizes the communication volume of query tensors and new caches. Based on TokenLake, the scheduler can schedule requests elastically by using existing techniques without considering prefix cache management. Evaluations on real-world workloads show that TokenLake can improve throughput by up to 2.6× and 2.0× and boost hit rate by 2.0× and 2.1×, compared to state-of-the-art cache-aware routing and cache-centric PD-disaggregation solutions, respectively.