Advanced Operating System Technologies

	Sergio Cittolin
   CERN - Switzerland

        Fabio Riccardi 
  CERN - Switzerland and
Dipartimento di Informatica e Sistemistica
dell' Universita` degli Studi di Napoli - Italy


        Sandro Vascotto
  CERN - Switzerland

                Abstract

Large distributed computing systems are foreseen to be the core part of the DAQ and Control system of the future LHC experiments. Neworks of thousands of processors, handling dataflows of several gigaBytes per second, with very strict timing constraints (microseconds), will become a common experience in the following years. Problems like distributyed scheduling, real-time communication protocols, failure-tolerance, distributed monitoring and debugging will have to be faced. A solid software infrastructure will be required to manage this very complicared environment, and at this moment neither CERN has the necessary expertise to build it, nor any similar commercial implementation exists.

Fortunately these problems are not unique to the particle and high energy physics experiments, and the current research work in the distributed systems field, especially in the distributed operating systems area, is trying to address many of the above mentioned issues. The world that we are going to face in the next ten years will be quite different and surely much more interconnected than the one we see now. Very ambitious projects exist, planning to link towns, nations and the world in a single "Data Highway". Teleconferencing, Video on Demend, Distributed Multimedia Applications are just a few examples of the very demanding tasks to which the computer industry is committing itself. This projects are triggering a great research effort in the distributed, real-time micro-kernel based operating systems field and in the software enginering areas.

The purpose of our group is to collect the outcame of these different research efforts, and to extabilish a working environment where the different ideas and techniques can be tested, evaluated and possibly extended, to address the requirements of a DAQ and Control System suitable for LHC.

Our work started in the second half of 1994, with a research agreement between CERN and Chorus Systemes (France), world leader in the micro-kernel OS technology. The Chorus OS is targeted to distributed real-time applications, and it can very efficiently support different "OS personalities" in the same environment, like Posix, UNIX, and a CORBA compliant distributed object architecture.

Projects are being set-up to verify the suitability of our work for LHC applications, we are building a scaled-down prototype of the DAQ system foreseen for the CMS experiment at LHC, where we will directly test our protocols and where we will be able to make measurements and benchmarks, guiding our development and allowing us to build an analytical model of the system, suitable for simulation and large scale verification.

Submitter:      Fabio Riccardi
Institution:    CERN
Address:        1211 Geneva 23, Switzerland
email:          fabio.riccardi@cern.ch
Tel:            + 41 22 767 3541
Fax:            + 41 22 767 3100

Topic Area:     DAQ and Triggering

Title:          Advanced Operating System Technologies

Authors:        Sergio Cittolin, CERN - Switzerland

                Fabio Riccardi, CERN - Switzerland and
                Dipartimento di Informatica e Sistemistica
                dell' Universita` degli Studi di Napoli - Italy

                Sandro Vascotto
                CERN - Switzerland