This project contains the workload traces used in the simulations for the paper entitled "Managing Overloaded Hosts for Dynamic Consolidation of Virtual Machines in Cloud Data Centers Under Quality of Service Constraints", which has been accepted to IEEE Transactions on Parallel and Distributed Systems.
The data have originally been provided as a part of the CoMon project, a monitoring infrastructure for PlanetLab: http://comon.cs.princeton.edu/.
The simulated host was assumed to be equipped with a quad-core CPU. The clock frequency of a single core of the host was set to 3 GHz, which according to the model introduced in the paper transforms into 12 GHz. These CPU characteristics correspond to a mid-range Amazon EC2 physical server type. The amount of the host’s memory was assumed to be enough for the Virtual Machines (VMs) allocated to the host. The CPU frequency of a VM was randomly set to one of the values approximately corresponding to the Amazon EC2 instance types: 1.7 GHz, 2 GHz, 2.4 GHz, and 3 GHz.
The data are in the comma-delimited CSV format, where the columns are:
Simulation ID - the index of a simulation, or a set of VMs assigned to the host
VM ID - the index of a VM within a simulation
Time Frame - the index of a time frame within a simulation
CPU Frequency - the CPU frequency of a VM in MHz
Utilization - the VM's CPU utilization as a percentage of the VM's CPU frequency
Copyright (C) 2012 Anton Beloglazov