Jul 2026· IEEE International Conference on Cloud Computing· pp. 78-88· 0 citations· 36 references
Abstract
Serverless computing has emerged as a promising paradigm for executing scientific workflows characterized by complex task dependencies, data-intensive operations, and high computational demands. However, most existing scheduling approaches assume homogeneous, single-region environments and primarily focus on isolated function execution. These assumptions overlook two critical challenges: (i) the inherent data-parallel nature of workflow tasks, and (ii) the heterogeneity of computing resources across geo-distributed serverless platforms. In this paper, we address these limitations by proposing a novel scheduling framework for geo-distributed serverless environments that explicitly models intra-function data parallelism, heterogeneous abstract resources, and regional concurrency constraints. We formulate a makespan minimization problem in which each function can either execute entirely on a single high-capacity resource or be partitioned across multiple heterogeneous lower-capacity resources, subject to region-specific concurrency limits.To solve this problem, we design a Deep Q-Network (DQN)-based scheduler augmented with two auxiliary heuristics. The first, Critical Workload First, prioritizes high-workload functions through an exhaustive split-deployment search over heterogeneous abstract serverless resources. The second, Load-Aware Heuristic, selects execution regions using a weighted load metric combined with penalty-based resource assignment. We evaluate our approach on five representative scientific workflows BWA, Montage, Inspiral, CyberShake, and SIPHT and using real-world round-trip time measurements from Azure Function deployments across three geo-distributed regions. Experimental results demonstrate that our DQN-based scheduler reduces makespan by up to 29.13% compared to state-of-the-art approaches.
As computing resources in cloud environments become increasingly abundant, executing complex scientific workflows on large-scale cloud infrastructure has become a standard practice. However, communication-intensive workflows face two fundamental bottlenecks. First, the lack of physical topology awareness often forces high-frequency interacting microservices to be placed on geographically distant nodes, which generates excessive cross-node communication overhead, leads to network load imbalance, and increases latency. Second, the prohibitive online computation time of conventional iterative scheduling algorithms further degrades response speed, making them unsuitable for real-time scenarios. To address these bottlenecks, this paper proposes AMSche, a framework for microservice deployment and task scheduling that is aware of both position and topology. The framework comprises two core mechanisms. The first mechanism, position-aware service deployment, colocates high-frequency interacting services on the same physical node based on communication affinity, thereby compressing cross-node communication overhead at the physical level. The second mechanism, topology-aware task scheduling, leverages online topology feature similarity mapping to instantly reuse historical scheduling plans, achieving scheduling decisions at the millisecond level. Extensive experiments on real-world scientific workflow datasets demonstrate that AMSche achieves an average improvement of 16.19% to 39.15% over existing baseline methods in comprehensive metrics including response time, total communication volume, and network load balance.
Hao Wei, Hailiang Chen, Jia-Nan Sun et al.· Fall Joint Computer Conferen...· 0 citations
Public LLM services serve diverse multi-agent applications with varying workflow dependencies and performance requirements. Requests generated by these applications often exhibit commonality and interdependence, yet current systems largely ignore such application-level structure. As a result, at the LLM engine cluster level, assigning requests to engines with the shortest queue can cause inefficient KV-cache transfers across GPUs. Using three representative multi-agent applications, we show that current scheduling methods miss opportunities to (a) improve performance through KV-cache reuse and reduced data transfer, and (b) increase goodput via batch management informed by workflow dependencies. Motivated by these observations, we propose a Workflow-Aware Scheduling system for Multi-Agent LLM systems (WaSMa) that incorporates cluster-and engine-level scheduling to optimize LLM request execution across GPU resources. Experimental results show that WaSMa reduces the P95 tail latency by up to 48% and improves goodput by up to 107% compared to existing methods.
Uttam Rao, Ali Zafar Sadiq, Hai-Ying Shen et al.· International Conference on...· 0 citations
This paper presents a novel simulation framework for characterizing the interplay between taskset granularity and system-level constraints and demonstrates that hybrid composition strategies, which dynamically balance taskset independence with execution grouping, can yield up to 3.8x throughput increase and a 14.9x reduction in network overhead.
Concurrent multi-agent workflows expose future dependencies and serving-state requirements while running on heterogeneous GPU pools with time-varying load, model residency, and resource availability. The logical workflow defines the required computation, whereas its physical scheduling units, model-lifecycle actions, resource ordering, and placement must be selected according to the observed pool state. We present a prediction-guided runtime that uses workflow forecasts to construct and optimize a physical execution graph. Predictor estimates device-specific activation latency, peak memory, and model-loading cost, then propagates these predictions through workflow dependencies to forecast activation readiness and future model demand. Constructor builds semantics-preserving fusion and model-lifecycle alternatives, while Scheduler jointly optimizes their selection, placement, and execution order based on the live pool state. Across a workload spanning three workflow scenarios on a heterogeneous GPU pool, our system reduces end-to-end makespan and overall p95 completion latency under burst arrivals by up to 36.8% and 25.9%, respectively, over state-of-the-art workflow schedulers. It also saves up to 24.63 GPU-s per completed session.
Jing-Hao Wang, Yi-Feng Zhang, Xiao Zhou et al.· 0 citations
A resource-aware optimization framework that dynamically selects the MPI process count and performs node- and NUMA-aware process placement and reduces task-sequence execution time and improves the evaluated resource-utilization metrics by more than 30%.
Wenxiao Wang, Zi-Bo Gao, Guoding Ji et al.· Journal of Intelligent Compu...· 0 citations
The Latency-Aware Adaptive Spotted Hyena Optimizer (LA-ASHO) is proposed, a novel metaheuristic scheduling framework grounded in the social hunting behaviour of spotted hyenas that achieves statistically significant reductions in workflow completion latency relative to established baseline schedulers such as; Min-Min, Heterogeneous Earliest Finish Time (HEFT).
Igiri C. G, Ejekwu Obunezi, Ujah Alechenu Israel· Journal of Artificial Intell...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
