Parallelism in DB Operations (cont)
Parallel nested loop join
- each outer tuple needs to examine each inner tuple
- but only if it could potentially join
- range/hash partitioning reduce partitions to consider
Parallel sort-merge join
- as noted above, parallel sort gives range partitioning
- merging partitioned tables has no parallelism (but is fast)
|
![[prev]](icon/prev.gif){kind=icon}
![[next]](icon/next.gif){kind=icon}