Process management inside ATLAS DAQ

Autor: D. Burckhart-Chromek, A. Ribeiro, I. Alexandrov, Helmut Wolters, D. Schweiger, Lorenzo Moneta, Dietrich Liko, S. Kolos, Marc Dobson, E. Badescu, Livio Mapelli, António Amorim, Luis G. Pedro, Y. F. Ryabov, Igor Soloviev, R. W. L. Jones, V. Roumiantsev, R. Hart, Levi Lucio, M. Nassiakou, Andrei Kazarov, Mihai Caprini, Mikhail Mineev, V. M. Kotov, P.Y. Duval
Přispěvatelé: Centre de Physique des Particules de Marseille (CPPM), Aix Marseille Université (AMU)-Institut National de Physique Nucléaire et de Physique des Particules du CNRS (IN2P3)-Centre National de la Recherche Scientifique (CNRS), ATLAS, Damoiseaux, Magali
Jazyk: angličtina
Rok vydání: 2001
Předmět:
Zdroj: IEEE Transactions on Nuclear Science
IEEE 2001 Nuclear Science Symposium (NSS) And Medical Imaging Conference (MIC)
IEEE 2001 Nuclear Science Symposium (NSS) And Medical Imaging Conference (MIC), Nov 2001, San Diego, United States. pp.2459-2462
Scopus-Elsevier
CIÊNCIAVITAE
ISSN: 0018-9499
Popis: The Process Management component of the online software of the future ATLAS experiment data acquisition system is presented. The purpose of the Process Manager is to perform basic job control of the software components of the data acquisition system. It is capable of starting, stopping and monitoring the status of those components on the data acquisition processors independent of the underlying operating system. Its architecture is designed on the basis of a server client model using CORBA based communication. The server part relies on C++ software agent objects acting as an interface between the local operating system and client applications. Some of the major design challenges of the software agents were to achieve the maximum degree of autonomy possible, to create processes aware of dynamic conditions in their environment and with the ability to determine corresponding actions. Issues such as the performance of the agents in terms of time needed for process creation and destruction, the scalability of the system taking into consideration the final ATLAS configuration and minimizing the use of hardware resources were also of critical importance. Besides the details given on the architecture and the implementation, we also present scalability and performance tests results of the Process Manager system.
Databáze: OpenAIRE