The ALICE Data Acquisition System

                Abstract

The first one consists of the acquisition and reduction of the huge volume of data produced by Pb-Pb interactions. The experiment will produce a low rate (50 Hz) of very large events (25 MBytes) resulting in an unprecedented volume of data of the order of 1 GBytes/sec after zero-suppression. A large computing facility will preprocess the events with the goal of reducing the volume of data recorded onto permanent storage.
The second one concerns a possible extension of the physics covered by ALICE which proposes to measure high-mass lepton pairs. This would result in an additional higher event rate superimposed to the previous one. For these events, the TPC which is producing 90 % of the data for Pb-Pb interactions will not be readout.
Finally, the ALICE experiment will take data with heavy-ion beams 4 weeks per year. The rest of the running period will be used to acquire data produced by pp interactions which produces less data. During this period of the year, the computing infrastructure will be used partly for data acquisition and partly for data analysis.

These different, and sometimes conflicting requirements require that the computing system is flexible enough to cope with them. In particular, a data acquisition of a new generation has therefore to be simulated, designed and developed. The efforts of the collaboration is therefore engaged in several directions.

A flexible simulation program has been designed to simulate the current requirements and to evaluate the relative merits of the possible architecture proposed for the data acquisition system and the different technologies existing to implement it.
The usage of point-to-point high speed links and of a switch is the most promising technique today. Different standards or technologies (ATM, HiPPI, Fibre Channel, SCI) could be used for the high-speed links and the switch which will transport the data and build the events. We have started an effort to study the applicability of Fibre Channel in this area and to prototype some of the DAQ components. In parallel, the different methods of switch-based event-building are studied.
This foreseen large data-acquisition system will require a distributed and flexible control system. Cooperation has started with the CERN ECP control group to define the user requirements to build a run control system.

Submitter's Name: VANDE VYVRE Pierre
Submitter's Institution: CERN - ECP Division
Address of Institution: 1211 GENEVA 23
Submitter's EMAIL address: PVV@VXCERN.CERN.CH
Submitter's Telephone number and/or FAX number: Tel (41-22) 767.49.45
                                                Fax (41-22) 767.54.90
Authors of the Paper each with Institution (and Experiment
Affiliation, if any):

J.Nelson, O.Villalobos Baillie (University of Birmingham)
E.Denes, A.Holba, G.Rubin, L.Szedrei (KFKI-RMKI, Budapest)
T.Kiss, Z.Meggyesi (Technical University of Budapest)
G.Harangozo, R.McLaren, H.G.Ritter, E.Van Der Bij, P.Vande Vyvre (CERN, Geneva)
B.Kolb, M.Purschke (GSI, Darmstadt)
G.Maron, M.Beldishevski (INFN, Legnaro)
B.Kvamme, B. Skaali, B.Wu (University of Oslo)
H.Beker (Universita "La Sapienza" and Sezione INFN, Roma)
(The ALICE Collaboration)