TRIERS: traffic burst oriented adaptive resource provisioning in cloud

A holistic solution for Traffic Burst Oriented Adaptive Resource Provisioning in Cloud (TRIERS for short) has been proposed to tackle the problem of maintaining performance scalability and service agility in the presence of sudden bursts of workload traffic. The TRIERS framework consists of three core components: workload classification by characterization, traffic prediction in class, and dynamic resource configuration and scheduling. Extensive experiments have been conducted to validate its practicality in terms of guarantee ratio (25.7% improved), resource utilization (5.6% improved) and total energy consumption (17.3% saved), indicating the effectiveness of TRIERS for adaptive resource provisioning in workload traffic burst.