抄録
This paper investigates the integration of Horizontal Pod Autoscaling (HPA) into containerized microservices architectures, focusing on optimizing performance, resource utilization, and operational costs. Building upon prior research on microservices grouping strategies, experiments were conducted using the Sock-Shop benchmark tool across low, medium, and high workload scenarios. Results reveal that HPA significantly enhances scalability, throughput, and latency—achieving up to 66% improved throughput and 32% reduced response times under certain grouping strategies. However, these advantages come with trade-offs, such as increased disk usage and operational complexity. This study provides a detailed analysis of HPA’s impact on dynamic environments and offers practical recommendations for balancing performance and cost in deployment strategies. Future research directions include exploring alternative scaling models, diverse workload impacts, serverless integration, and operational simplifications.
https://ror.org/036rp1748 
