A Wireless Architecture For Performance Monitoring And Visualization On PDA Devices Claudio Zunino Fabrizio Lamberti Andrea Sanna Bartolomeo Montrucchio Dipartimento di Automatica e Informatica, Politecnico di Torino, corso Duca degli Abruzzi 24, I-10129 Torino (Italy) {c.zunino,lamberti,sanna,montru}@polito.it ABSTRACT Information Visualization requires the merging of data visualization methods, computer graphics solutions and graphics interfaces design. In performance monitoring applications, the use of Information Visualization is essential to guarantee an effective and easy consultation of information. The employment of PDA devices could allow an effective monitoring thanks to the use of wireless technologies but their poor resources have limited their application until now. A wireless PDA-based architecture for cluster performance monitoring and visualization is presented. In this paper the issue of efficient data presentation on PDA devices has been addressed and compared to a traditional Web PC-based approach. The effectiveness of the proposed architecture in a real environment is shown. Keywords: information visualization, PDA devices, wireless architecture, cluster performance monitoring, Web-based application. 1. INTRODUCTION In a broad spectrum of disciplines a large amount of information has to be visualized. Information Visualization, is an important sub-discipline within the field of scientific visualization and it focuses on visual mechanisms designed to communicate clearly to the user the structure of information and improve the cost of access to large data repositories. Information Visualization enables people to deal with all of this information by taking advantage of visual perception capabilities of the human being. By presenting information more graphically it is possible for the human brain to use more of its perceptual system in initially processing information, rather than immediately relying entirely on the cognitive system. Some of the most important papers in the field are collected in [1]. An attractive field of research in Information Visualization is monitoring and management of parallel systems by means of advanced graphics user interfaces. Moreover, a further challenge is the employment of PDA devices in accomplishing these tasks. The monitoring operation is devoted to obtain a performance analysis; for instance, the user can know in real time the percentage of CPU used at each node, the amount of memory used for a specified application, as well as the network loading, the space available on the hard disk, and many other useful information. All these statistics allow to evaluate the behavior both of a specified application and of the hardware composing the system. On the other hand, management operations can include bootstrap and shutdown of every node as well as more complex commands such as task migration to accomplish load balancing, or security option setting. Information has to be collected and presented to the user in an intuitive way; a set of suitable icons should be used in order to show CPU usage, memory and disk occupation, number of tasks and of logged users, network traffic, and so on. The monitoring architecture should be easily scalable (it has to be possible adding and removing nodes without affecting the monitoring system structure), network topology independent, and reliable. In this work the performance monitoring task is addressed through a wireless architecture. The proposed architecture has been developed in order to make performance monitoring information visualization possible even on mobile devices such as Personal Digital Assistant (PDA) based systems. PDA devices, that were originally designed as personal organizers, are are today effective pocket computers; in fact, by means of additional software, their functionalities are virtually unlimited. Moreover, a comparison with a traditional Web-based visualization architecture is provided. Section 2 reviews the main visualization system devoted to cluster monitoring and summarizes nowadays PDA technical characteristics. Section 3 presents the proposed architecture in terms of the infrastructure for performance information collection and of the visualization interface for monitoring data presentation to the user. Section 4 presents the detailed implementation. Section 5 analyzes advantages and drawbacks of the proposed solution. 2. BACKGROUND In this section, the main works known in the literature aimed to monitor and manage parallel/distributed architectures are reviewed (in subsection Cluster monitoring) and the most important characteristics of PDAs are highlighted (in subsection PDA devices). Cluster monitoring-information collection and visualization Performance data visualization techniques have evolved from 2D static to immersive virtual environments (a survey can be found in [2]). This section reviews the main works known in the literature focusing on solutions proposed for cluster monitoring. A first solution to collect and display information on the status of a Unix-like machine is to use the Rstatd daemon. Rstatd provides information to remote hosts via the SunRPC [3] (Sun Remote Procedure Call protocol); statistics provided by Rstatd are shown in Table 1.