SCSQ Super Computer Stream Query processor

1
SCSQ Super Computer Stream Query processor

• Research issues
• Develop and investigate methods that ensure
  scalable query execution for increasing
• stream volume
• number of stream sources
• computational cost of query operators
• Optimize execution of continuous queries in the
  target environment taking into account both
  computation and communication time
• Optimal utilization of the BlueGene
• Optimal utilization of the communi-cation
  subsystems inside and outside BlueGene

• LOFAR is the worlds 1st all-software radio
  telescope
• The telescope consists of antenna fields,
  producing 22 Tbps raw data streams
• We implement SCSQ, a Data Stream Management
  System for processing of data streams from the
  LOFAR antennas.
• SCSQ features
• Users search and analyze contents of data streams
  using continuous queries
• Execution of such continuous queries in a
  massively parallel hardware environment

SCSQ Erik ZeitlerTore RischUDBL In
collaboration with LOFAR and LOIS Supported by
LOFAR
Central processing facilities feature a 12000
CPU IBM BlueGene. This is the target hardware
environment for SCSQ.
SCSQ architecture. Brown boxes are SCSQ
components.
Scientists submit continuous queries to SCSQ and
receive query result streams
CQ
Result
Antenna signals
Result
CQ
Department of Information Technology Box
337 SE-751 05 Uppsala Sweden
Forskningens dag 2005-04-04 Erik Zeitler