Last update: Apr 1, 2000
Speaker: Igor SfiligoiThe main kind of data in KLOE are physical events, both raw and reconstructed. Their number and size is several orders of degrees higher than of the other types of data, i.e. configuration constants and condition variables.
The number of events foreseen in KLOE is estimated to reach 200 billion, with a storage request of nearly 1 Pb. To manage these numbers, for long term storage the events are grouped in files of reasonable size (500 Mb at present) and managed using this metric. Moreover, seen the required space, only a small amount of these files can reside on the secondary storage.
The access to the events is however totally transparent to the final user. Using a c-library he is able to access any event independently of its location; it makes no difference for the user weather the event is stored somewhere on the tertiary storage, in a memory buffer or somewhere in the middle.
The system is based on a set of in-house developed daemons, interfaced to industry standard products for the RDBMS and archiving (IBM DB2 and ADSM at present).
The talk will cover a description of the system architecture as well as performance and robustness of the system.
|| | Home | Bulletins | Committees | Scientific Program | Docs by topics | Social Event | Conference Location | Secretariat ||