# Mell: Memory-Efficient Large Language Model Serving via Multi-GPU KV Cache Management

arXiv: https://arxiv.org/abs/2501.06709 · HTML: https://mlsanity.com/paper/2501.06709/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in Serving systems and request scheduling, one of 12 directions in Efficient LLM inference and serving.
- Builds on 30 of its 41 references that are in the map.
- First posted on arXiv 12 Jan 2025.

## Placed in

- Efficient LLM inference and serving / [Serving systems and request scheduling](https://mlsanity.com/inference/serving-systems-and-request-scheduling/)

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

- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [H2O: Heavy-Hitter Oracle for Efficient Generative Inference of Large Language Models](https://mlsanity.com/paper/2306.14048/) (2023)
- [KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache](https://mlsanity.com/paper/2402.02750/) (2024)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Serving](https://mlsanity.com/paper/2401.09670/) (2024)
- [FlexGen: High-Throughput Generative Inference of Large Language Models with a Single GPU](https://mlsanity.com/paper/2303.06865/) (2023)
- [Splitwise: Efficient generative LLM inference using phase splitting](https://mlsanity.com/paper/2311.18677/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [Taming Throughput-Latency Tradeoff in LLM Inference with Sarathi-Serve](https://mlsanity.com/paper/2403.02310/) (2024)
- [Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs](https://mlsanity.com/paper/2310.01801/) (2023)
- [LMSYS-Chat-1M: A Large-Scale Real-World LLM Conversation Dataset](https://mlsanity.com/paper/2309.11998/) (2023)
- [Scissorhands: Exploiting the Persistence of Importance Hypothesis for LLM KV Cache Compression at Test Time](https://mlsanity.com/paper/2305.17118/) (2023)
- [DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale](https://mlsanity.com/paper/2207.00032/) (2022)
- [InfiniGen: Efficient Generative Inference of Large Language Models with Dynamic KV Cache Management](https://mlsanity.com/paper/2406.19707/) (2024)
- [LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models](https://mlsanity.com/paper/2309.12307/) (2023)
- [Llumnix: Dynamic Scheduling for Large Language Model Serving](https://mlsanity.com/paper/2406.03243/) (2024)
- [Cost-Efficient Large Language Model Serving for Multi-turn Conversations with CachedAttention](https://mlsanity.com/paper/2403.19708/) (2024)
- [Keyformer: KV Cache Reduction through Key Tokens Selection for Efficient Generative Inference](https://mlsanity.com/paper/2403.09054/) (2024)
- [Inference without Interference: Disaggregate LLM Inference for Mixed Downstream Workloads](https://mlsanity.com/paper/2401.11181/) (2024)
- [Response Length Perception and Sequence Scheduling: An LLM-Empowered LLM Inference Pipeline](https://mlsanity.com/paper/2305.13144/) (2023)
- [ServerlessLLM: Low-Latency Serverless Inference for Large Language Models](https://mlsanity.com/paper/2401.14351/) (2024)
- [TurboTransformers: An Efficient GPU Serving System For Transformer Models](https://mlsanity.com/paper/2010.05680/) (2020)
- [Characterization of Large Language Model Development in the Datacenter](https://mlsanity.com/paper/2403.07648/) (2024)
- [Galaxy: A Resource-Efficient Collaborative Edge AI System for In-situ Transformer Inference](https://mlsanity.com/paper/2405.17245/) (2024)
- [OTAS: An Elastic Transformer Serving System via Token Adaptation](https://mlsanity.com/paper/2401.05031/) (2024)
- [Efficient Online Strategies for Renting Servers in the Cloud](https://mlsanity.com/paper/1408.4156/) (2014)

## Built on by (most cited first, 8 of 8)

- [Prism: Cost-Efficient Multi-LLM Serving via GPU Memory Ballooning](https://mlsanity.com/paper/2505.04021/) (2025)
- [PPAI: Enabling Personalized LLM Agent Interoperability for Collaborative Edge Intelligence](https://mlsanity.com/paper/2605.18067/) (2026)
- [ForkKV: Scaling Multi-LoRA Agent Serving via Copy-on-Write Disaggregated KV Cache](https://mlsanity.com/paper/2604.06370/) (2026)
- [Crossflow: Prefill-Decode Elasticity for Agentic LLM Serving](https://mlsanity.com/paper/2609.27085/) (2026)
- [Functional Cache Grafting: Robust and Rapid Code-Policy Synthesis for Embodied Agents](https://mlsanity.com/paper/2606.13097/) (2026)
- [MATCH: Modulating Attention via In-Context Retrieval for Long-Context Transformers](https://mlsanity.com/paper/2606.29844/) (2026)
- [Energy-Aware Scheduling for Serverless LLM Serving on Shared GPUs](https://mlsanity.com/paper/2606.30391/) (2026)
- [LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs](https://mlsanity.com/paper/2608.03036/) (2026)

## Abstract (arXiv)

Serving large language models (LLMs) for massive users is challenged by the significant memory footprint of the transient state, known as the key-value (KV) cache, which scales with sequence length and number of requests. Instead of renting or buying more expensive GPUs, the load imbalance of the KV cache across GPUs, coupled with recent advances in inter-GPU communication, provides an opportunity to serve more requests via request migration. However, high migration overhead and unpredictable request patterns make it challenging. Therefore, this paper proposes MELL, a memory-efficient LLM serving system via multi-GPU KV cache management. It saves the number of GPUs needed in the system by considering the dynamic KV cache load and the costly request migration. Specifically, we first develop an adaptive request migration mechanism to balance the computational and communication overheads and adapt to diverse resource conditions. Then, we design an online algorithm tailored to a multi-LLM request and multi-GPU scheduling problem with migration enabled. It aims to minimise the required GPUs while limiting the number of migrations. Finally, we implement a prototype of MELL and demonstrate that it reduces the number of GPUs by 31% and increases the GPU utilization by 43% at most compared to existing LLM serving systems.