Quick Review (cont.)

• Sorting and Hashing

  • Both are memory intensive
  • Major concerns
    • Merge efficiency & memory manangement (S)
    • Hash table overflow & skewing (H)

• Disk Access

  • Indices usage
  • Physical database design

Previous slide

Next slide

Back to first slide

View graphic version

Sorting problems “solutions”:

Merge efficiency- forecasting (predict largest key), large cluster size, using max fan-ins (# of runs that can be merged at once).

Hashing problems “solutions”:

Overflow- avoidance (partition data according to fan-out (# of partition files made)), resolution (assume overflow won’t occur & resort to avoidance when it does).