Autonomic Request Management Algorithms for Geographically Distributed Internet-Based Systems

Abstract

Supporting Web-based services through geographical distributed clusters of servers is a common solution to the increasing volume and variability of modern traffic. These architectures pose interesting challenges to request management strategies where the most important goal is not to achieve maximum performance, but to guarantee stable and robust results. In this paper, we propose novel request management algorithms that are based on autonomic principles that is, on loose collaboration among the closest nodes and no knowledge about the global system state. Experimental evaluation shows that our autonomic-enhanced algorithms can guarantee robust performance in a variety of settings and reduce standard deviations of the response times with respect to existing request management algorithms.

Control-theoretical load-balancing for cloud applications with brownout

“…Often load-balancing policies consider web server systems as a target [11,26], where one of the most important result is to bound the maximum response time that the clients are exposed to [19]. Load-balancing strategies can be guided by many different purposes, for example geographical [2,33], driven by the electricity price to reduce the datacenter operation cost [15], or specifically designed for cloud applications [5,23,24].…”

“…For this, we use the user-perceived stability σ u [2]. 1 http://cvxopt.org/ This metric refers to the variation of performance as observed by the users, and it is measured as the standard deviation of response times.…”

A Model and Algorithm for Self-Adaptation in Service-oriented Systems

“…Quitadamo et al [4] demonstrate a knowledge-network driven approach to service selection and aggregation. Andreolini et al [12] exploit load trends for autonomic request forwarding between geographically distributed systems.…”

Adaptive control of cluster-based Web systems using neuro-fuzzy models

“…The solutions encountered include the application of a more efficient server in the service, proper scheduling of HTTP requests on the input to the Web server (AlSa'deh and Yahya, 2008;Harchol-Balter et al, 2003;Zatwarnicki, 2010), scheduling and admission control in the Web server (Borzemski and Suchacka, 2010;Elnikety et al, 2004;Lee et al, 2004;Quan and Chung, 2005;Wei et al, 2005;Wei and Xu, 2006), the application of a locally distributed cluster-based Web system (Borzemski and Zatwarnicki, 2003;Cardellini et al, 2001;Cherkasova and Karlsson, 2001;Pai et al, 1998), and the application of globally distributed Web server clusters (Andreolini et al, 2008;Borzemski et al, 2007). Forecasting data transfer times in the Internet is of considerable significance as well (Borzemski, 2006).…”