Towards efficient deployment of cloud applications through dynamic reverse proxy optimization

With the increase of users and the deployment requests, the issue of dynamic deployment in PaaS becomes prominent. Different approaches of application deployment have been deeply discussed, but the issues like fast response to a large number of concurrent deployment requests are rarely focused. In this work, we extend Nginx as a dynamic reverse proxy to support dynamically remote configuration for better elasticity of cloud applications in PaaS, and then further optimize it for improving performance under a large number of concurrent configuration requests. Three optimization approaches are proposed: Batch Request Committing (BRC), Batch File Processing (BFP) and In Memory Configuration (IMC). We give a detailed implementation of each method, and a qualitative analysis on three optimization approaches has been made. Finally, a series of experiments are presented to validate the optimization effect. The experiment results show that the maximum throughput per second has increased significantly, and the average response time of each request has decreased dramatically.