The Community for Technology Leaders
2015 IEEE World Congress on Services (SERVICES) (2015)
New York City, NY, USA
June 27, 2015 to July 2, 2015
ISBN: 978-1-4673-7274-9
pp: 31-38
There has been a growing effort in decreasing energy consumption of large-scale cloud data centers via maximization of host-level utilization and load balancing techniques. However, with the recent introduction of Container as a Service (CaaS) by cloud providers, maximizing the utilization at virtual machine (VM) level becomes essential. To this end, this paper focuses on finding efficient virtual machine sizes for hosting containers in such a way that the workload is executed with minimum wastage of resources on VM level. Suitable VM sizes for containers are calculated, and application tasks are grouped and clustered based on their usage patterns obtained from historical data. Furthermore, tasks are mapped to containers and containers are hosted on their associated VM types. We analyzed clouds' trace logs from Google cluster and consider the cloud workload variances, which is crucial for testing and validating our proposed solutions. Experimental results showed up to 7.55% improvement in the average energy consumption compared to baseline scenarios where the virtual machine sizes are fixed. In addition, comparing to the baseline scenarios, the total number of VMs instantiated for hosting the containers is also improved by 68% on average.
Virtual machining, Containers, Energy consumption, Google, Servers, Cloud computing, Clustering algorithms

S. F. Piraghaj, A. V. Dastjerdi, R. N. Calheiros and R. Buyya, "Efficient Virtual Machine Sizing for Hosting Containers as a Service (SERVICES 2015)," 2015 IEEE World Congress on Services (SERVICES), New York City, NY, USA, 2015, pp. 31-38.
90 ms
(Ver 3.3 (11022016))