TERN: A Self-Adjusting Thermal Model for Dynamic Resource Provisioning in Data Centers

Dynamic resource provisioning becomes a practical approach to achieving high thermal and energy efficiency, improving scalability, and optimizing reliability for e-commercial applications running in modern data centers. In this paper, we propose a self-adjusting model called TERN to predict thermal behaviors of hardware resources for client sessions. Our TERN contains two major components: (1) a resource utilization model being responsible for estimating hardware usage based on the number of running client transactions, and (2) a thermal model that discovers correlation between resource utilization and their temperatures. TERN is conducive to predicting thermal trends of diverse workload conditions with a changing transaction mix. We apply the TPC-W benchmark to characterize the resource demands of each type of transactions. Then, we conduct a thermal profiling study of the benchmark with various transaction mixes. TERN judiciously adjusts the models to maintain prediction accuracy for dynamically changing request patterns. Experimental results show that TERN provides a simple yet powerful solution for resource provisioning in thermal-aware data centers where exist rapidly changing workload conditions.