Resource Scheduling for Real-Time Analytical Workflow Services in the Cloud

Nowadays, most data analytical applications comprise of multiple tasks, which can be represented as workflow in nature. Some of data analytical applications, the data requests arrived continuously, such as fraud detection application and order application. In general, such workflow applications have a rigid requirement in relation to response time. When running the analytical workflow in a cloud platform, one of the critical questions which arise is how to provision resources so that the monetary cost can be reduced while guaranteeing system throughput. In this paper, we use queueing network theory to address this challenge. First, we present the performance analytic model for the elastic analytical workflows based on queueing network theory. Then, we design a resource provision strategy to determine the number of virtual machines for hosting components of the applications with throughput guarantee. Both real experiments and simulation experiments using the real workload traces data show that our proposed approach provides a simple yet powerful solution to provision resources for analytical workflows under dynamic workload conditions.