Publication No 40709

Author(s)

Mashaly, M.

Title

Managing Load Balancing, Energy Efficiency and Performance of Cloud Data Centers with Service Level Agreement Guarantees

Abstract

Cloud Data Centers (CDC) have become a major network resource with a high energy consumption. In this thesis two fundamental energy-efficiency algorithms are modeled mathematically: (1) server consolidation by state-dependent control of compute server activation/deactivations, considering server activation overhead, low power server sleep modes, and Dynamic Voltage and Frequency Scaling (DVFS), (2) two dynamic load balancing algorithms through virtual machine (VM) migrations between local or distant server groups of CDCs. Modeling approach (1) is based on a novel multi-level state hystereses by a Finite State Machine (FSM); modeling approach (2) is based on two novel load balancing (LB) mechanisms between CDC server groups through Virtual Machine (VM) migrations, both under Service Level Agreement (SLA) constraints. The mathematical analyses are based on multi-dimensional Markov Chains (MC), from which all relevant performance quantities are derived. The analytical methods are validated by computer simulations based on the OMNeT++ software tool in cases (1) and (2) or by experiments and measurements on a small EMC2 server cluster.

Year

2017

Reference entry

Mashaly, M.
Managing Load Balancing, Energy Efficiency and Performance of Cloud Data Centers with Service Level Agreement Guarantees
Dissertation, Stuttgart, Informatik, Elektrotechnik und Informationstechnik, 2017

BibTex file

Download  [BIBTEX]

Full Text

Download  [PDF]