The default scheduler of Kubernetes, the state-of-the-art container orchestrator, uses fast, local placement decisions. Unfortunately, this design leads to resource fragmentation, reduced cluster usage, and overprovisioning. External solvers can compute global placement plans, but enforcing these plans in upstream clusters is hard. Kubernetes provides no native cross-node preemption, uncoordinated concurrent scheduling leads to inconsistencies, and replacing the default scheduler would sever deployments from upstream cycles. We present OPSche, an open-source Kubernetes Scheduling Framework plugin where external solvers can drive cluster-wide placement decisions in concert with the default scheduler. OPSche atomically validates and enforces solver-produced plans through coordinated framework hooks and supports three trigger modes: scheduling-failure, periodic, and stable-queue -- resp. triggered when a workload cannot be placed, at fixed time intervals, when the set of pending workloads stabilises. Each mode has a blocking variant for a finer tuning of placement quality, latency, and disruption. We pair OPSche with a constraint-based optimisation solver, showing its feasibility across a broad set of cluster configurations and reporting improvements of resource usage by up to 3.0% and scheduling latency by more than a second.
A comprehensive review of Kubernetes scheduling strategies published between January 2023 and January 2026 is presented and a multi-dimensional taxonomy is established that categorizes scheduling approaches based on common objectives, modification methods, optimization methodologies, targeted workloads, evaluation methods, scheduling scopes, and performance metrics.
Mohammed Alhakimi, R. Latip· Computers· 0 citations
Efficient workload scheduling is central to the performance, scalability, and cost-effectiveness of modern clouds. In the most popular cloud platform today, Kubernetes (K8s), the scheduler can be extended with multiple scheduling plugins, enabling fine-grained control over scheduling decisions. However, cloud providers often lack clear insights into the trade-offs of individual placement strategies and resort to combining multiple plugins without a clear rationale. This uncertainty creates a gap in understanding which scheduling configuration yields optimal results for a given application or traffic pattern under specific infrastructure conditions. Thus, this paper investigates the implications of different scheduling strategies in K8s by leveraging KubeTwin (KT), a high-fidelity simulator for the K8s platform designed for reproducible experimentation. This work extended KT with advanced scheduling capabilities to evaluate the impact of single and combined scheduling plugins across diverse workloads and infrastructure conditions, focusing on key aspects such as performance, fairness, and resource utilization. The results highlight that Node-affinity and Diktyo single-purpose plugins consistently outperform multi-objective scheduling combinations, achieving the lowest deployment costs (below 50 price/day), shortest queue times (under 50 ms), and fastest response times (below 200 ms). In contrast, combined strategies, such as Balanced, result in significantly higher costs and higher latency. This study shows that combining multiple optimization criteria within a single scheduling strategy often degrades performance, whereas targeted, single-objective plugins deliver robust and predictable behavior across heterogeneous infrastructures and workload patterns.
José Santos, D. Borsatti, Walter Cerroni et al.· IEEE Conference on Network S...· 1 citation
Modern cloud-native applications increasingly rely on Kubernetes to orchestrate large-scale, heterogeneous workloads. However, experimenting with its cluster-level mechanisms in a scalable way, such as its diverse scheduling policies, remains difficult in production environments due to operational costs and limited observability. This demonstration showcases KubeTwin, a practical framework that creates and maintains a live Digital Twin of a Kubernetes cluster. KubeTwin enables the modeling of multi-cluster infrastructures, including edge and cloud nodes, heterogeneous resource capacities, and network latencies, allowing for safe experimentation, debugging, and what-if analysis without disrupting production workloads. Through an interactive demonstration, we show how researchers and developers can leverage KubeTwin to (i) assess the performance of workloads under diverse traffic patterns and (ii) evaluate diverse scheduling strategies under different infrastructure scenarios. The demo highlights how Kubernetes Digital Twins can bridge the gap between simulation and real-world experimentation, providing a powerful tool for both operational decision-making and research prototyping.
José Santos, D. Borsatti, Walter Cerroni et al.· IEEE Conference on Network S...· 0 citations
Wide Nextflow fan-outs on shared Slurm clusters can submit tens of thousands of short tasks. Deployment settings route them through individual jobs, arrays, or nested schedulers inside enclosing allocations. These settings determine workflow turnaround and RPC volume, a shared cost that can degrade scheduler responsiveness. Existing comparisons evaluate whole workflow systems, while per-task queueing metrics cannot span architectures that dispatch inside existing allocations. We contribute a reproducible measurement protocol and benchmark harness. A clean-start clock begins before backend startup or allocation requests, placing architecturally different backends on a common time axis. Per-user Slurm sdiag counters attribute request count as the primary RPC demand measure and controller processing time as sensitivity context, separate from cluster-wide state. We apply the method to Slurm native dispatch, Slurm job arrays, HyperQueue, and Flux on the shared ASU Phoenix production cluster and a single-user Dev cluster. On Phoenix, every aggregation strategy improves both objectives relative to native dispatch; Flux has the lowest RPC demand, while HyperQueue's fastest median walltime is not stable across replicates. On Dev, arrays and Flux improve walltime, while HyperQueue trades the lowest RPC demand for the slowest completion. The method lets an HPC site compare deployment strategies using both user-visible performance and scheduler impact, then select the fastest strategy within its own RPC-demand limit.
Nil Tianchen Mu, William Dizon, Glen Otero et al.· 0 citations
Serverless platforms maintain a global view of function invocations and resource utilization, yet existing systems largely restrict CPU scheduling decisions to the operating system scheduler. This paper presents a serverless platform-driven CPU load balancing framework that enables the control plane to directly influence CPU scheduling through a custom Linux scheduler built on SchedExt(SCX). The proposed scheduler introduces configurable scheduling domains and a shared interface that allows the control plane to assign functions to domains based on their historical inter-arrival times. Within each domain, a single-queue load-balancing strategy combined with a virtual-time prioritization policy improves task placement while reducing interference from busy-polling tasks. Results show that an eight-domain configuration achieves the best trade-off, reducing system energy consumption by approximately 15% while increasing invocation cost by only 5%. Under heavily loaded workloads, the proposed scheduler also reduces request latency by up to 50% compared to the default Linux Completely Fair Scheduler (CFS). These results demonstrate that exposing CPU scheduling decisions to the serverless control plane can improve both energy efficiency and workload performance while preserving scheduling flexibility.
Scientific computing is increasingly shifting from monolithic applications to coupled simulation-AI workflows composed of highly heterogeneous tasks with diverse hardware, scale, and runtime requirements. As these workflows scale to leadership-class systems, the resulting extreme ensemble sizes and task variability can create orchestration bottlenecks. System-level schedulers are often configured for limited throughput, while workflow tools face scalability issues due to rigid control-plane topologies and static scheduling heuristics. We introduce EnsembleLauncher, a recursively hierarchical workflow orchestrator for exascale systems, featuring a fully decentralized control plane and a programmable scheduling policy interface. On the Aurora supercomputer, EnsembleLauncher successfully scales to the entire machine with up to eight million serial tasks, outperforming state-of-the-art tools by more than four times. Additionally, we implement a programmable scheduling interface and demonstrate a significant impact of scheduling policies on resource utilization for high-variance ensembles and active learning pipelines representative of modern coupled simulation-AI workflows.
Harikrishna Tummalapalli, Christine M. Simpson, Riccardo Balin et al.· 0 citations