Full text
| |
| Author(s): |
Rodrigo Augusto Cardoso da Silva
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Campinas, SP. |
| Institution: | Universidade Estadual de Campinas (UNICAMP). Instituto de Computação |
| Defense date: | 2015-12-15 |
| Examining board members: |
Nelson Luis Saldanha da Fonseca;
Antonio Marinho Pilla Barcellos;
Eduardo Candido Xavier
|
| Advisor: | Nelson Luis Saldanha da Fonseca |
| Abstract | |
The high energy consumption is a significant problem in cloud data center operations. Such consumption can be minimized by means of proper management of data center resources. This dissertation studies the energy consumption in cloud data centers and proposes two new algorithms for the minimization of energy consumption. The Topology-aware Virtual Machine Placement (TAVMP) algorithm deals with the placement of virtual machines considering hierarchical data center network topologies and workload demands modeled as virtual machine groups. Results show that it is possible to reduce blocking ratio of requests to allocate virtual machines while maintaining acceptable levels of energy consumption. Moreover, a load balancing algorithm to promote energy savings is proposed. The Topology-aware Virtual Machine Selection (TAVMS) algorithm chooses sets of groups of virtual machines to be migrated to other data centers in a distributed data center scenario considering hierarchical data center network topologies. Results show that by using this algorithm it is possible to obtain relevant energy savings. Results obtained using discrete event simulations indicate that the consideration of the network topology, workload demands as groups of virtual machines and their traffic demands is fundamental to achieve energy savings in a distributed data center scenario (AU) | |
| FAPESP's process: | 14/01154-4 - Energy-aware load balancing in distributed data centers |
| Grantee: | Rodrigo Augusto Cardoso da Silva |
| Support Opportunities: | Scholarships in Brazil - Master |