소프트 랜딩 스케일러: 웹 트래픽 폭주에 대한 효율적 대응을 위한 적응적 자원 할당 전략

Abstract

This research paper addresses managing sudden web traffic increases in cloud computing, emphasizing effective resource allocation. Traffic surges can overwhelm servers, leading to service instability and dissatisfaction. The proposed

solution, the Soft Landing Scaler (SLS), dynamically adjusts resources based on traffic fluctuations using a Kubernetes- based architecture. SLS is designed for optimal resource efficiency and adaptability, maintaining satisfactory response

times. The study analyzes traffic patterns such as Sharp Increase then Exponential Decrease (SIED), and Sharp Increase then Linear Decrease (SILD), demonstrating SLS's performance. Results show improved resource efficiency and user response times, highlighting SLS's effectiveness in handling diverse traffic surges. The study contributes to the field by presenting an adaptive scaling system focused on response times, using real-world traffic data, and emphasizing buffer resources and scaling size limits in downscaling strategies.