Skip to content
Conference

Cost-Aware Auto-Scaling for Serverless Computing Under Dynamic Workloads

Aug 2026 · International Workshop on Artificial Intelligence and Cognition · pp. 85-90 · 0 citations · 16 references

Abstract

The popularity of serverless computing in cloud computing has increased as it has fine-grained billing model, automatic scaling, and low operation overhead. Nevertheless, available serverless auto-scaling solutions are performance-oriented and do not know anything about the cost of execution, which commonly results in the ineffective use of resources and higher cloud costs when it comes to dynamically load-related workloads and bursts. The current paper presents a proposal of an auto-scaling framework of serverless computing that is cost-conscious and reduces both execution cost and quality-of-service (QoS) constraints. The proposed solution incorporates real-time workload tracing. Other factors also considered such as execution cost modeling and QoS-aware scaling choices to dynamically adapt concurrency of functions, memory allocation and pre-warming policies. An estimation of the impact of billing is developed mathematically as a cost model and scaling actions are calculated by solving a constrained optimization problem minimizing cost subject to latency requirements. Hours of experimentation under both steady and bursty workloads has shown that the proposed framework is much less expensive to run than the default provider-driven auto-scaling by up to 40-45 percent, and has similar latency, throughput and SLA adherence. The proposed strategy also works well in terms of the cold start frequency resulting in a better latency stability and user experience. These findings indicate the success of applying the concept of cost intelligence to serverless auto-scaling and give practical recommendations on how to implement cost-effective serverless applications in the cloud.

View source

Similar papers

Cold start performance optimization in serverless computing platforms

This paper analyzes the main influencing factors related to the formation mechanism and optimization path of cold starts, focusing on code download, instance startup, runtime initialization, and user code initialization, and summarizes methods such as container reuse, snapshot recovery, warm-up and elastic scaling, int...

Chen-Ye Kang · 0 citations
Book Open access Sep 2026

QMScaler: A QoS-Constrained Resource-Efficient Microservice Autoscaling Framework for Edge Environments

QMScaler is proposed, a QoS-driven microservice horizontal scaling framework based on Monte Carlo Tree Search (MCTS) that aims to minimize the number of container instances while meeting the QoS requirements of multiple application functions.

Tian-Yang Zheng, Peng-Fei Yang, Zhe Xu et al. · 0 citations
Open access 2026

WORKLOADAWARE: A Resource Allocation and Consolidation Technique for Heterogeneous Clouds

—Datacenters are the core of cloud computing services that consume a significant amount of energy and negatively affect our environment. These problems can be solved using efficient resource management techniques, such as resource allocation and consolidation methods. This paper models the resource allocation problem a...

Muhammad Zakarya, A. Khan, Lee Gillam et al. · 0 citations
Open access Oct 2026

EVALUATION OF SERVERLESS COLD START OPTIMIZATION TECHNIQUES

Cold start latency remains one of the main challenges in serverless computing, particularly for Function-as-a-Service (FaaS) applications operating under dynamic workloads. Runtime initialization, dependency loading, and resource allocation increase response latency and reduce system performance. Although numerous cold...

O. Kisseleva, К. К. Абдраманов, N. Vitaliy et al. · 0 citations
Open access Sep 2026

EASE: Resource-aware Query Scheduling across Heterogeneous Cloud Compute Services

Modern cloud platforms provide diverse compute services, including virtual machines, Function-as-a-Service, and Query-as-a-Service, each offering unique trade-offs in performance, elasticity, and cost. While these services collectively cover the diverse needs of OLAP workloads, existing systems typically rely on a sing...

Wen-Bo Li, Hao-Qiong Bian, Chao Zhang et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.