I love coding because it simplifies many things in life. During my spare time, I always work to improve my coding abilities and hope to contribute to open-source projects.
[email protected]
0953-866-708
Expecting work region: Taipei
High energy consumption of data centers is an important issue which is widely discussed. Dynamic consolidation of Virtual Machine (VM) can reduce energy consumption by concentrating the workload of active hosts. However, VM migrations would cause some cost and consolidation might bring more resource demands and lead to violate service level agreements (SLA) between cloud computing providers and users.
We proposed a VM allocation mechanism by using the requirement history of VMs to calculate standard deviations (STD), the mechanism can predict future resource requirements of each VM and reallocate VMs into the server under the QoS requirements. When some servers overload in a datacenter, suitable VMs are selected from these servers and migrated to appropriate servers to meet the requirement of SLA. On the other hand, when some servers underload, all VMs are migrated out from underload server, switched to sleep mode. Compared to the heuristics proposed in previous studies, the proposed method could greatly decrease number of migrations、SLA violations and execution time.
I love coding because it simplifies many things in life. During my spare time, I always work to improve my coding abilities and hope to contribute to open-source projects.
[email protected]
0953-866-708
Expecting work region: Taipei
High energy consumption of data centers is an important issue which is widely discussed. Dynamic consolidation of Virtual Machine (VM) can reduce energy consumption by concentrating the workload of active hosts. However, VM migrations would cause some cost and consolidation might bring more resource demands and lead to violate service level agreements (SLA) between cloud computing providers and users.
We proposed a VM allocation mechanism by using the requirement history of VMs to calculate standard deviations (STD), the mechanism can predict future resource requirements of each VM and reallocate VMs into the server under the QoS requirements. When some servers overload in a datacenter, suitable VMs are selected from these servers and migrated to appropriate servers to meet the requirement of SLA. On the other hand, when some servers underload, all VMs are migrated out from underload server, switched to sleep mode. Compared to the heuristics proposed in previous studies, the proposed method could greatly decrease number of migrations、SLA violations and execution time.