Cargando…

Implementation and performance analysis of the LHCb LFC replica using Oracle streams technology

The presentation will describe the architecture and the deployment of the LHCb read-only File Catalogue for the LHC Computing Grid (LFC) replica implemented at the Italian INFN National Centre for Telematics and Informatics (CNAF), and evaluate a series of tests on the LFC with replica. The LHCb com...

Descripción completa

Detalles Bibliográficos
Autores principales: Düllmann, D, Carbone, A, Martelli, B, Peco, G, Bonifazzi, F, Da Fonte Perez, E, Baranowski, Z, Vagnoni, V
Lenguaje:eng
Publicado: 2007
Materias:
Acceso en línea:http://cds.cern.ch/record/1120784
Descripción
Sumario:The presentation will describe the architecture and the deployment of the LHCb read-only File Catalogue for the LHC Computing Grid (LFC) replica implemented at the Italian INFN National Centre for Telematics and Informatics (CNAF), and evaluate a series of tests on the LFC with replica. The LHCb computing model foresees the replication of the central LFC database in every Tier-1, in order to assure more scalability and fault tolerance to LHCb applications Scientific data intensive applications use a large collection of files for storing data. In particular, as regards the HEP community, data generated by large detectors will be managed and stored using databases. The intensive access to information stored in databases by the Grid computing applications requires a distributed database replication in order to guarantee the scalability and, in case of failure, redundancy. Besides the results of the tests will be an important reference for all the Grid users This talk will describe the replica implementation of LHCb LFC database that is foreseen to be done to each LHCb Tier-1 site. The LFC database replication is realized by means of Oracle Streams technology, a high-speed tool that allows synchronization and full consistency of replicated database. The stream technology takes care of propagating information from one database to another. In order to verify the requirements asked by the LHCb Collaboration a series of tests has been performed. The tests are designed to evaluate performances, delay on the propagation of the streams and the scalability of LFC. The tests show the robustness of the replica implementation with performance going beyond the experiment's requirements. In our tests we also take into account these issues verifying the ability of the LFC to retry a connection if the Oracle server becomes unavailable, as well as the ability of the LHCb applications to choose an LFC replica and if this replica fails. The LHCb LFC replica implementation will allow to the LHCb users to perform data analysis with high performance, scalability and fault tolerance. The results of this experience will be useful for many other EGEE applications based on Oracle databases. Applications need to be aware of the underlying replicated environment in order to take advantage of the added fault tolerance and scalability.