Power-Aware Replica Placement in Tree Networks with Multiple Servers per Client

نویسندگان

  • Guillaume Aupy
  • Anne Benoit
  • Matthieu Journault
  • Yves Robert
چکیده

In this paper, we revisit the well-studied problem of replica placement in tree networks. Rather than minimizing the number of servers needed to serve all client requests, we aim at minimizing the total power consumed by these servers. In addition, we use the most general (and powerful) server assignment policy, where the requests of a client can be served by multiple servers located in the (unique) path from this client to the root of the tree. We consider multi-modal servers that can operate at a set of discrete speeds, using the dynamic voltage and frequency scaling (DVFS) technique. The optimization problem is to determine an optimal location of the servers in the tree, as well as the speed at which each server is operated. A major result is the NP-completeness of this problem, to be contrasted with the minimization of the number of servers, which has polynomial complexity. Another important contribution is the formulation of a Mixed Integer Linear Program (MILP) for the problem, together with the design of several polynomialtime heuristics. We assess the efficiency of these heuristics by simulation. For mid-size instances (up to 30 nodes in the tree), we evaluate their absolute performance by comparison with the optimal solution (obtained via the MILP). The most efficient heuristics provide satisfactory results, within 20% of the optimal solution. Key-words: replica placement, tree network, multiple servers, power-aware algorithm, complexity, discrete speeds. ∗ Ecole Normale Supérieure de Lyon, LIP, CNRS, INRIA, Université de Lyon, France † Institut Universitaire de France, France ‡ University of Tennessee Knoxville, USA Placement de répliques dans des réseaux en forme d’arbre sous contraintes de puissance avec plusieurs serveurs par client Résumé : Dans ce papier, nous revisitons le problème de placement de répliques dans des réseaux en forme d’arbre. Plutôt que de minimiser le nombre de serveurs requis pour satisfaire toutes les requêtes des clients, nous visons à minimiser la puissance totale consommée par ces serveurs. Nous utilisons la politique d’assignation la plus générale où les requêtes d’un client peuvent être servies sur plusieurs serveurs, tous placés sur le chemin entre ce client et la racine de l’arbre. Nous considérons des serveurs multi-modaux qui peuvent travailler à plusieurs vitesses (ensemble discret de vitesses), en employant la technique du DVFS (Dynamic Voltage and Frequency Scaling). Le problème d’optimisation revient à choisir un placement optimal des serveurs sur les nœuds de l’arbre, ainsi qu’à déterminer leurs vitesses d’exécution. Alors que le problème de minimisation du nombre de serveurs pouvait être résolu en temps polynomial, nous montrons que ce nouveau problème est NP-complet. Nous formulons toutefois un programme linéaire en nombres entiers permettant de résoudre ce problème de façon optimale (mais en un temps exponentiel), ainsi que plusieures heuristiques polynomiales. Par de multiples simulations, nous avons étudié ces heuristiques, en les comparant pour des instances de tailles moyenne (jusqu’à 30 nœuds) à la solution optimale obtenue par le programme linéaire. Les heuristiques les plus efficaces sont en moyenne à un facteur de 30% de la solution optimale. Mots-clés : placement de répliques, arbres, serveurs multiples, puissance, complexité, vitesses discrètes. Power-aware replica placement in tree networks 3

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Improve Replica Placement in Content Distribution Networks with Hybrid Technique

The increased using of the Internet and its accelerated growth leads to reduced network bandwidth and the capacity of servers; therefore, the quality of Internet services is unacceptable for users while the efficient and effective delivery of content on the web has an important role to play in improving performance. Content distribution networks were introduced to address this issue. Replicatin...

متن کامل

Comparison of Access Policies for Replica Placement in Tree Networks

In this paper, we discuss and compare several policies to place replicas in tree networks subject to server capacity. The client requests are known beforehand, while the number and location of the servers are to be determined. The standard approach in the literature is to enforce that all requests of a client be served by a single server in the tree (Single). One major contribution of this pape...

متن کامل

Impact of QoS on Replica Placement in Tree Networks

This paper discusses and compares several policies to place replicas in tree networks, subject to server capacity and QoS constraints. The client requests are known beforehand, while the number and location of the servers are to be determined. We study three strategies. The first two strategies assign each client to a unique server while the third allows requests of a client to be processed by ...

متن کامل

Optimal Replica Placement in Tree Networks with QoS and Bandwidth Constraints and the Closest Allocation Policy

This paper deals with the replica placement problem on fully homogeneous tree networks known as the Replica Placement optimization problem. The client requests are known beforehand, while the number and location of the servers are to be determined. We investigate the latter problem using the Closest access policy when adding QoS and bandwidth constraints. We propose an optimal algorithm in two ...

متن کامل

Optimal Closest Policy with QoS and Bandwidth Constraints for Placing Replicas in Tree Networks

This paper deals with the replica placement problem on fully homogeneous tree networks known as the Replica Placement optimization problem. The client requests are known beforehand, while the number and location of the servers are to be determined. We investigate the latter problem using the Closest access policy when adding QoS and bandwidth constraints. We propose an optimal algorithm in two ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2014