Re: Index on partial field? No - so then?

Based on note read project is migration from IMS

If you have well designed IMS databases then migrating to a relational scheme should be very straight forward. Generally each segment in an IMS database become a table with dependent segments taking the key of the root and the dependent segment together to form a primary key. A secondary index can then be created on any column ( dbd defined field ) that has a secondary IMS index on it.

The original note wanted to know about indexing a partial column. While Oracle can not directly do this there are ways to accomplish this in Oracle. If an index exists on this column either single column or first column or multi-column index then the SQL parameter 'like' and its associated wildcard characters may fill the bill. Where they will not, then by adding another column to the table and creating both an before insert and before update trigger on the table to populate the new column the application barely needs to know it exists. It does help for the 'select'-ing program to reference this new column.

Just a couple of ideas from someone who spend a few years as an IMS DBA. For those who are unfamiliar with IMS it can handle a load. Many of the largest banks in the world use IMS fast path data bases. Oracle is a heck of a lot easier to set up an application in, but once set up it is hard to beat the speed with which IMS can process a transaction. Tradeoffs, tradeoffs, ... Received on Thu Feb 06 1997 - 00:00:00 CST