Loading...
1 results
Search Results
Now showing 1 - 1 of 1
- Simulation and Performance Evaluation of a Distributed Virtual-Machine Placement and Migration Approach for Management of Cloud Computing Resources Using CloudSim PlusPublication . Filho, Manoel Campos da Silva; Freire, Mário Marques; Monteiro, Claudio de CastroCloud computing services are widely available over the Internet. One of the main reasons, which motivates its large-scale adoption, is the easiness for users to hire a wide range of services in the pay-per-use model, where you only pay for consumed services and resources. Such services predominantly rely on server virtualization technologies such as virtual machines (VMs). They enable sharing of computing resources between different customers, in an isolated way, looking to ensure the correct functioning of services, security and data privacy from each customer. Despite cloud computing technologies are consolidated, there are many issues and technological challenges to maintain and manage the computing infrastructure of providers and ensure availability, stability, reliability and quality of service to customers. One of the providers goals is the efficient distribution of VMs between available servers, targeting: load balancing, fault tolerance, reduction of network traffic, efficient use of limited computational resources and, mainly, reduction of energy consumption and costs. Trying to find solutions for efficient allocation of computing resources in the cloud, this thesis presents: (i) the issues with computer simulation tools to carry out research in the field; (ii) the challenges of managing computing resources in the cloud; (iii) the cloud computing simulation framework called CloudSim Plus, for creation of more accurate and reliable simulation experiments; and (iv) the development of a proposal for placement and migration of VMs. CloudSim Plus applies several software engineering principles to increase assurance of correct operation of the tool and created simulations, giving higher confidence in produced results. This tool is used in this thesis for development and assessment of a distributed VM placement and migration proposal. The large scale of cloud computing infrastructures, and the inherent complexity of algorithms for optimizing the allocation of computing resources, requires proposals with distributed architecture, looking to balance the processing load of such algorithms to reduce the search time for sub-optimal solutions. CloudSim Plus aims both to (i) solve reliability and accuracy issues of its precursor Cloud- Sim and (ii) be the state-of-the-art tool for general-purpose cloud computing simulation. The VM placement and migrating proposal is based on the Modern Portfolio Theory (MPT), widely applied in the financial market for creation of efficient investment portfolios, where the investor capital is allocated in different assets (such as stocks) to balance risk and return. Since (i) the risk in cloud infrastructures can be defined as the variation in the use or availability of computing resources for contracting customers and (ii) the return, being the proper allocation and use of such resources by those customers, it can be drawn a parallel between cloud computing services and the financial market. The proposal represents a proof of concept for creating computer simulations with CloudSim Plus, using a distributed, fault-tolerant architecture with reduced computational complexity compared to previous proposals.