Distributed Object-Oriented Applications Supervision N. Cottin, J. Gaber, O. Baala, M. Wack Nathanael.Cottin, Jaafar.Gaber, Oumaya.Baala, Maxime.Wack @utbm.fr University of Technology of Belfort-Montbeliard 90010 Belfort, France Abstract Distributed object-oriented computing allows efficient use of the Network Of Workstations (NOW) paradigm. How- ever, the underlying middlewares used to develop and de- ploy such applications do not provide developers with any standard supervision mechanism so that they know exactly what happens during their applications execution. This paper analyzes distributed CORBA and JAVA-based applications to point out functional and management su- pervision information which has to be gathered from the objects. Developers will use this information to improve the Quality of Service (QoS) of their distributed object-oriented applications (DOA). Keywords : Distributed computing, supervision, QoS, JAVA, CORBA. 1 Introduction We consider distributed applications from a supervision perspective. Supervision is based on observation whereas management uses observed information to perform modi- fications on the supervised entities. Thus, managing dis- tributed object-oriented applications (DOA) requires super- vision mechanisms. Management systems rely on ad-hoc supervision indicators to take QoS improvement decisions. Management tools developement is a complex process. De- velopers tasks can be simplified by using a supervision API. In this paper, we present a supervision API and propose su- pervision indicators relevant to a management system. We focus on CORBA and JAVA-based applications [1] to show how this API could be implemented. The rest of the paper is organized as follows : next section provides an overview of the work related to the distributed applications QoS improvement issue. Then we present the background material necessary to seize our study. This includes distributed applications representation and super- vision implementation. These two sections lead to intro- duce a supervision API inherited by distributed objects as well as supervision indicators based on this API. We finally conclude by presenting our work in progress following our study. 2 Related work Some projects have proposed ways to improve dis- tributed CORBA-based applications QoS in terms of re- sponse time and overhead. Among them, QUARTZ, LSS and DOMS are representative of this research area. How- ever, QUARTZ and LSS are dedicated to special-purpose applications such as media and service-based applications whereas DOMS is a general-purpose QoS improvement system. Another way of supervising distributed applica- tions is to directly instrument the ORB [2]. However, this solution is ORB-dependant and not portable. 2.1 QUARTZ Quartz [3] is a generic QoS description environment integrated with CORBA developped at Trinity College Dublin. It is used to guarantee that CORBA-based dis- tributed applications meet specified QoS criteria for control and transfert of streaming media. Quartz interprets QoS parameters required by applications to allocate corresponding resources at system-level. It uses application and system-level filters to translate QoS con- straints and allocate resources according to various proto- cols such as RSVP, TCP/IP and ATM. A C++ prototype us- ing Iona Orbix has been developped. It uses Windows NT real time capabilities as well as the audio and video stream- ing mechanism standardized by the OMG [4]. 2.2 LSS The IT Research Center of Montreal has designed a ser- vice facility called Load Sharing Service (LSS) [5]. This distributed CORBA-based system is used for managing ob- jects calls when multiple objects provide the same service type.