Multi-Tenant Service Performance Model And Performance Evaluation

In order to achieve service level agreements (SLAs) and maintain the equipment application (application server and database server) with high performance and utilization rate, multi-tenant application service system is facing a serious deployment optimization and performance optimization problem, according to the "load distribution, admission control, resource control". And accurately predicting "each interactive response time which is sent from the client server of the multi-tenant services", can solve this problem well.This task is particularly challenging due to the fact that the mixes of tenants with different business scale and operating characteristics and the interaction among the concurrently running queries have a great impact on the response time of queries in the multi-tenant applications system and an accurate model needs to capture them. What’s more, the client server, application server and database server is often a virtual machine which is deployed in cloud computing environment and its load, data and physical resources can often change. According to the dynamic change of load, data and physical resource allocation, the establishment of an effective, high accuracy, good robustness "multi-tenant services application performance model" is still a problem to be solved. In this paper, our goal is to build such a performance model for the interactions of multi-tenant using an experiment-driven modeling approach.We use a Bayesian approach and build novel Gaussian models that take into account a variety of factors that influence the response time of each interaction that sent from the different tenant respectively in the multi-tenant environments(detail in section 2). We experimentally demonstrate that our models are accurate and effective which have an average prediction error of 12.6% in the worst case.