Kbase P111784: What are impacts of manually removing progress_recid from non-unique indices?
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
12/13/2005 |
|
Status: Unverified
GOAL:
What are impacts of manually removing progress_recid from non-unique indices?
GOAL:
Why does Data server migration utility add progress_recid column to non-unique indices?
GOAL:
Do we recommend manually removing progress_recid from non unique indexes?
FACT(s) (Environment):
Oracle DataServer
FIX:
Oracle data server migration utility adds progress_recid column to non unique indice to make these indices unique.
When we submit queries that use non-unique indexes in 4GL application, we also add progress-recid to the query (when present) so that we have a unique value to find the record by, since the keys themselves in the index may not be unique. For example, if you have a "EACH customer USE-INDEX name" type of query in the application, the dataserver will have to add the ORDER BY clause to the SQL sent to ORACLE to guarantee the ordering of the result-set, and we will also add the progress_recid to the ORDER BY clause. If the column is not in the index, then ORACLE will have to do some sorting to get the records in the desired order, as a result you could see slower performance. As such, we do not recommend removing progress_recid from non unique indexes.