Aug 2026· Proceedings of the ACM SIGCOMM 2026 Conference· pp. 16-29· 0 citations· 50 references
TL;DR
DualPath is an inference system that breaks this bottleneck by introducing dual-path KV-Cache loading and enables a novel storage-to-decode path, in which the KV-Cache is loaded into decoding engines and then efficiently transferred to prefill engines via RDMA over the compute network.
Abstract
The performance of multi-turn, agentic LLM inference is increasingly dominated by KV-Cache storage I/O rather than computation. In prevalent disaggregated architectures, loading the massive KV-Cache from external storage creates a fundamental imbalance: storage NICs on prefill engines become bandwidth-saturated, while those on decoding engines remain idle. This asymmetry severely constrains overall system throughput. We present DualPath, an inference system that breaks this bottleneck by introducing dual-path KV-Cache loading. Beyond the traditional storage-to-prefill path, DualPath enables a novel storage-to-decode path, in which the KV-Cache is loaded into decoding engines and then efficiently transferred to prefill engines via RDMA over the compute network. DualPath combines this optimized data path — which inherently avoids network congestion and avoids interference with latency-critical model execution communications — with a global scheduler that dynamically balances load across prefill and decode engines. Our evaluation on three models with production agentic workloads demonstrates that DualPath improves offline inference throughput by up to 1.87x on our in-house inference system. It can also improve online serving throughput by an average factor of 1.96x without violating SLO.
Disaggregating the prefill and decoding stages of large language model (LLM) inference into two separate sets of nodes is widely adopted in today's LLM serving systems. However, such an architecture poses significant challenges for self-hosted LLM deployments on rented cloud instances, since transferring enormous key-value (KV) caches between disaggregated nodes can easily saturate the limited inter-node network bandwidth. In this paper, we propose to mitigate the network bottleneck by selectively transferring essential KV cache entries across the two stages. There are two challenges to achieve selective KV cache transfer, i.e., accurate KV selection during the prefill stage, and efficient KV fetching during the decoding stage. To address these challenges, we design SmartGen, a KV cache transfer engine that allows seamless disaggregated LLM inference with three data transfer paths. Specifically, we leverage 1) a profile-based proactive transfer path to identify and push essential KV cache entries to the decoding node during the prefill stage, 2) a parallel on-demand transfer path to simultaneously fetch remote and local KV cache entries during the decoding stage, and 3) a speculative transfer path to finally deliver all KV caches to the decoding node. Experimental results show that SmartGen reduces time-to-second-token by up to 4.3x compared with the typical full KV cache transfer approach while offering comparable subsequent decoding performance and accuracy.
Xuchuan Luo, Jiacheng Shen, Xin Wang et al.· 0 citations
This work proposes Turbo, a first-of-its-kind in-network aggregation system that accelerates long-context inference by offloading query broadcast and attention aggregation to switches and introduces a rolling forward scheme that propagates states to enable cross-stage updates.
Ying Wan, Yuchen Xu, Chuwen Zhang et al.· Proceedings of the ACM SIGCO...· 0 citations
This work argues that future inference infrastructure should allow decoupling of compute and KV Cache storage across cloud and datacenters, and proposes a vision for an Internet for the KV Cache, with KV Cache management working as a content-distribution system.
Siddhant Ray, Nick Feamster, Junchen Jiang· 0 citations
Prefix caching has become a key technique for LLM serving, and nowadays the reusable KVCache contents are often hosted on distributed servers. For long-context LLM inferences with high cache hit ratio, cross-server KVCache transmission has become an emerging performance bottleneck; such network-intensive LLM inferences are increasingly prevalent in the coming era of agentic AI. However, existing LLM inference engines are essentially compute-centric; we find that they are highly inefficient when serving such workloads due to compute-stage service blocking and ignorance of KVCache-transfer cost. To efficiently serve network-intensive LLM inferences, in this paper, we design Sanic, an optimized LLM engine that treats KVCache transmission as a first-class citizen. Viewing KVCache loading and computation as equally-significant stages, Sanic decouples their service control and allows each stage to progress autonomously in an asynchronous manner, thereby improving the overall resource utilization. Moreover, when scheduling competing LLM inferences, Sanic treats the KVCache loading delay as an independent factor in service cost modeling, which is more accurate and can yield better scheduling decisions. Our testbed experiments with diverse benchmarks show that, Sanic can substantially enhance the service efficiency of network-intensive LLM inferences, improving the SLO-attainment by up to 61.67%.
Weiye Wang, Chen Chen, Junxue Zhang et al.· Asia-Pacific Workshop on Net...· 0 citations
Disaggregated LLM inference separates the prefill and decode phases across GPU pools, generating massive KV-cache transfers. Because these transfers last hundreds of milliseconds to seconds, they behave as mega elephant flows that dominate link bandwidth utilization. In this regime, stateless ECMP can perform poorly: hash collisions may overload one spine link while leaving others idle, stretching transfer times by seconds. Yet this same persistence makes coordination practical. Since these flows are long-lived, even lightweight one-to-all coordination can be amortized over their lifetime. We present ARK, a distributed elephant-flow path reservation mechanism. ARK coordinates senders to choose source ports whose hashes map concurrent flows onto distinct spines, without requiring switch changes or receiver-side packet reordering. Packet-level RDMA simulations show that ARK reduces mean and P95 FCT by up to 27.3% and 34.0% under moderate load, and further reduces mean TTFT by up to 12.9%.
Hung-Chun Lin, Ting-Wei Hsu, Chung-En Ho et al.· 0 citations
Deploying large language models (LLMs) on edge nodes enables low-latency and privacy-preserving inference, but faces severe resource constraints under high-concurrence workloads. While existing inference systems leverage intranode key–value (KV) caching to improve efficiency, they largely neglect the unique complexities of multinode edge environments. Specifically, reactive KV cache eviction policies suffer from temporal uncertainty, often discarding reusable KV caches prematurely, while the tight coupling between request scheduling and cache placement often leads to myopic decisions that exacerbate load imbalance and resource contention. To address these challenges, we propose a dynamic block-level paradigm that treats KV blocks as the fundamental units for caching and scheduling, enabling dynamic sharing, generation, and eviction of arbitrary-length prefixes. We present complete modeling of the spatiotemporal coupling between scheduling and caching under block-level granularity, capturing intricate interactions overlooked by prior work. Based on this model, we design an online joint optimization algorithm, which applies to general edge LLM serving scenarios. The algorithm decouples spatiotemporal dependencies via randomized rounding over per-slot subproblems, achieving a balance between real-time responsiveness and long-term system efficiency. Theoretical analysis establishes high-probability near-optimality guarantees, and extensive experiments show that our method reduces the average time to first token (TTFT) by up to 54.02% over existing baselines.
Xishuo Li, Wei Jiao, Jun He et al.· IEEE Internet of Things Jour...· 0 citations