SORT

For the SORT operator, the following input arguments are shown:

Order columns

An ordered list of columns that are to be sorted (that is, the sort keys).

Uniqueness

A flag that is true if duplicates are to be eliminated.

Estimated rows

The estimated number of rows that will be produced when the sort is finished.

Estimated row width

The width, in bytes, of a single row in the sort buffer.

Aggregation mode

The group by operation may be divided into multiple steps. The aggregation mode describes the level of aggregation performed at each step:

• Complete - Aggregation is done in a single pass.
• Partial - Aggregations are done on chunks of input rows.
• Intermediate - Partial aggregation results are processed to generate consolidated partial results.
• Final - Partial aggregation results are processed and any final calculations performed to generate the aggregation result.

Intra-partition sort type

The type of shared sort that is being used:

• Shared - One sort is created and subagents are synchronized during insertion into the sort. When the sort is completed, subagents use a parallel scan to read the sort.
• Replicated - This sort is used where all subagents are required to read every row of the sort. One sort is created and subagents are synchronized during insertion into the sort. When the sort is completed, each subagent reads the entire sort.
• Partitioned - A separate sort is created for each subagent. The subagents hash on the sort partition columns to determine into which sort they should insert a row. This partitions the data by value. When the sort is completed, each subagent reads from one of the sort partitions.
• Round-robin - A separate sort is created for each subagent. A round-robin clock type algorithm is used to determine into which sort a row should be inserted. This means that during the insert phase, subagents take turns being responsible for inserting into each of the sorts. When the sort is completed, each subagent reads from one of the sort partitions.

(This item is displayed only for intra-partition parallelism or full parallelism (intra-partition and inter-partition)).

Intra-partition columns

The columns that intra-partition parallelism has used to partition data between subagents when the value of Intra-partition sort type is set to Partitioned. (This item is displayed only for intra-partition parallelism or full parallelism (intra-partition and inter-partition)).

[ Top of Page | Table of Contents | Glossary | Index ]

[ DB2 List of Books | Search the DB2 Books ]