[prev] 21 [next]

Data Storage in PDBs (cont)

Data-partitioning strategies for one table:
  • round-robin partitioning
    • cycle through nodes, new tuple added on "next" node
  • hash partitioning
    • use hash value to determine which node and page
  • range partitioning
    • ranges of attr values are assigned to processors

Storing data on multiple nodes maximises chance for parallel data access