Data Declustering

• Issues

  • data blocks with spatio-temporal extents
  • non-uniform data distribution
  • large number of disks

• Maximize disk parallelism

  • assign similar data blocks to different disks