Cloud Workload Clustering

With countless businesses and millions of customers dependent upon cloud infrastructure, cloud resource manage-ment is more critical now than ever. Workloads like VMs, Databases and micro services use cloud resources and their usage data is monitored by the cloud service providers. With the cloud workload data, workloads can be categorized based on their usage of CPU, Memory and Disk I/O Operations. Clustering the workload data based on these categories will infer an understanding of the characteristics of workloads, usage of resources and efficient allocation of resources to the cloud service providers. Multiple clustering methods are compared and analysed thereby helping in smooth scaling without impacting the QoS (Quality of Service) of existing users. In past, Disk IO operations weren't CPU bottlenecked due to their low Disk IOPS, but with rising IOPS in storage devices, it can be seen if this still holds true.