Cloud computing has emerged as a new computational paradigm capable of managing largescale IT infrastructure. However, most of the existing cloud infrastructures are not efficiently operated, and resource overprovisioning is an emerging issue.
Due to the time-varying requirements of virtual resources, physical platforms might be inefficiently used, resulting in additional operational costs. Migration techniques were proposed to improve physical resource utilization, e.g. consolidating virtual resources on physical ones. Prior work driven by energy aware and load balancing objectives are often restricted to a single migration technique. This thesis presents an optimized migration model for virtual machines based on current and future states of physical resource usage while considering multiple migration techniques. The future usage is based on workload prediction. Our model aims at minimizing the operational cost when sharing the underlying infrastructure.
The experimentations carried out in this work demonstrate the ability of our model to perform better physical resources sharing compared to other models based on one single migration technique and with no prediction techniques. With our model, we managed to reduce the operational cost by 16% on average compared to an existing solution.
| Date | 11 Jul 2017 |
|---|
| Original language | French |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Mohamed Cheriet (Supervisor) & Kim Khoa Nguyen (Co-supervisor) |
|---|
Labidi, A. (Author),
Cheriet (Supervisor) &
Nguyen (Co-supervisor),
11 Jul 2017Student thesis: Master's thesis › Master in Engineering: Engineering