Predictive Auto-Scaling Framework for Preventing Server Overload in High-Traffic Web Applications

Abstract

Many web systems, such as university portals, banking platforms, and ticket booking websites, sometimes show a server busy error or experience server crashes when many users try to access the system at the same time. During peak hours server is not able to handle all requests properly. Because of that users get slow response and sometimes service interruptions also happens. So users face problems and trust on system also getting reduced.In this work a predictive auto scaling framework is designed to reduce server busy issues and make website more stable when many users are using at same time. Idea is simple users should not face server problems even if traffic suddenly increases. Here a cloud based approach is used. Number of users and incoming requests are monitored continuously. When traffic increase is predicted extra virtual servers are prepared before and workload is distributed across different machines so load can be handled better. As a result system performance is improved and chances of server crash are reduced. Overall heavy traffic can be handled in better way and users will get better service.