Initial Design of PEP production Data Base

Paper: 405
Session: C (poster)
Presenter: Ye, Yanlin, Peking University, Beijing
Keywords: data bases, ODBMS's, object-oriented methods, world-wide collaboration

Initial Design of PEP production Data Base

Y. Ye, J. Ying, D. Liu

Peking University, Bejing 100871

S. Sorensen, D. Morrison, V. Perevoztchikov, Y. Zhao, G. Han

University of Tennessee

for the PHENIX Collaboration


PEP is a fully distributed event processing tool for PHENIX experiment.
It consists of a large set of individual processes performing tasks. All
these processes can run across a WAN and communicate
to each other via a message passing system PVM. Since the system will handle
very large data size and will incorporate many kind of data operators which
maight run under very different conditions, it is necessary to keep all useful
information, related to data files, data structures and operation tasks etc.,
into a data base. PEP has a central controller which keeps track of the state
of each process and allocates resources. Therefore it is possible to take
all interested information just from the controller and to build an "unified"
data base for PEP.

The initial design consists of objects like PEP_TASK, PEP_DOP, PEP_JOB,
PEP_FILE, PHX_RUN, DATA_STRUCT etc, which are related to each other
according to an information model. The model has initially been
implemented on relational database ORACLE located at BNL and on a
local object data base POSTGRES. the Study, expansion and test of the
model are underway.