Realistic Workload Generation for Cloud Data Centers
Furkan Koltuk, Ece Güran Schmidt · 2020
This paper proposes a new method for creating synthetic workload traces in accordance with the distribution and time characteristics of a given actual workload trace. To this end, we first find the distribution that fits to the actual workload trace, then rearrange the random samples that are generated from this distribution such that the final synthetic trace has time characteristics that are similar to the actual trace. We evaluate our method using real virtual machine and task request traces of Azure and Google cloud data centers. Our method enables generating synthetic traces that can be used for a more realistic evaluation of cloud data centers.centers.