Data Storage in PDBs (cont)
Assume that partitioning is based on one attribute
Data-partitioning strategies for one table on n nodes:
- round-robin, hash-based, range-based
Round-robin partitioning
- cycle through nodes, new tuple added on "next" node
- e.g. i th tuple is placed on (i mod n)th node
- balances load on nodes; no help for querying
|
![[prev]](icon/prev.gif){kind=icon}
![[next]](icon/next.gif){kind=icon}