Archived project

Ubicacion de Maquinas Virtuales Multi-Objetivo con Acuerdo de Nivel de Servicio

0 new
0 new
0 new
0 new

Project log

Benjamin Baran
added 2 research items
Nowadays, cloud computing providers offer idle resources through an auction-based system in order to maximize resource utilization and economical revenue. Cloud computing consumers have the opportunity to take advantage of the resources offered at very low spot price in exchange for lower reliability in the provision of these resources. In this context, the Spot Price Prediction (SPP) is a well studied problem mainly formulated as a time series prediction, with particularities of auction-based cloud markets. This work presents a comparative evaluation of three different well-known prediction algorithms, applied for the first time to the SPP problem, against a state-of-the-art Three-Layer Perceptron (TLP) algorithm. In order to measure the accuracy of the evaluated algorithms, the following error metrics were considered: (1) Mean-Squared Error (2) Maximum Positive Error and (3) Mean Positive Error. Experimental results indicate that the Support Vector Poly Kernel Regression (SMOReg) algorithm outperforms other evaluated algorithms for the SPP problem, improving probabilities of obtaining resources in a highly dynamic spot market.
Cloud computing datacenters provide millions of virtual machines in actual cloud markets. In this context, Virtual Machine Placement (VMP) is one of the most challenging problems in cloud infrastructure management, considering the large number of possible optimization criteria and different formulations that could be studied. Considering the on-demand model of cloud computing, the VMP problem should be solved dynamically to efficiently attend typical workload of modern applications. This work proposes a taxonomy in order to understand possible challenges for Cloud Service Providers (CSPs) in dynamic environments, based on the most relevant dynamic parameters studied so far in the VMP literature. Based on the proposed taxonomy, several unexplored environments have been identified. To further study those research opportunities, sample workload traces for each particular environment are required; therefore, basic examples illustrate a preliminary work on dynamic workload trace generation.
Fabio Lopez-Pires
added 3 research items
The process of selecting which virtual machines will be placed (i.e. executed) in the physical machines available in a Datacenter is known as virtual machine placement problem. This work proposes for the first time a formulation of the problem, with a multi-objective approach, of the main objective functions studied so far as mono-objective in the state of the art. Also it is proposed a multi-objective memetic algorithm for solving the proposed problem. The validity of the proposed formulation is checked by comparing experimental results of the proposed algorithm with a brute force algorithm. Finally it is experimentally verified the scalability of the used meta-heuristic.
La virtualización permite usar dinámicamente los recursos de un Centro de Datos, según sean los requerimientos específicos. En este contexto, se denomina Problema de Ubicación de Máquinas Virtuales al proceso de seleccionar la máquina física más adecuada para crear (i.e., ejecutar) una máquina virtual requerida por un usuario. En este trabajo se propone una taxonomía de las soluciones al problema en cuestión, basada en los trabajos más relevantes del área, publicados en los últimos años en la biblioteca de la IEEE con el objetivo de guiar futuras investigaciones del problema. Finalmente son propuestos algunos temas de investigación todavía abiertos.
The process of selecting which virtual machines should be located (i.e. executed) at each physical machine of a Data center is known as Virtual Machine Placement - VMP. This work proposes for the first time a multi-objective formulation of the VMP considering Service Level Agreement. A novel multiobjective memetic algorithm is also proposed to solve the formulated multi-objective problem. This proposal is validated comparing experimental results of the proposed algorithm with a brute force exhaustive search algorithm. Simulations prove the correctness of the proposed memetic algorithm and its scalability considering different experimental scenarios.