indexes

[David Abramowitz]

There are several things you can do to help performance: Reorganize the file using the KEYFILE(*FILE) parameter if the physical file is indexed. For sequential access, use the SEQUENTIAL(*YES nnn) parameter of the OVRDBF command. The higher the number(nnn) the better up to about 512k characters. Create a vector index using SQL. The access times will surprise you. Dave

[Diane Mueller]

Dave, Thanks for your post, you've given me something to research on. I just want to understand this, once the index LTRindex was created for file LTRdtap, and if the file has never been reorganized, would this be why I see 0 for the i/o count for LTRindex when viewing the job. Since the LTRindex is listed with open files, it is not being used?? Diane

[Diane Mueller]

Our system has a file with over 8 million records, we are having a performance issue on this file - (sometimes).... In testing I have created an index over this large file, although when I watch the performance on the job, I notice the i/o count reads: LRTdtap (physical file-key on code only ) - I/O count...4057 LRTindex (index over cust# & code) I/O count....0 What I need to know is, if the open files list my index, but the I/O count shows 0, is the job actually using the index LRTindex? Should I not expect to see an i/o count of something more than 0 if it is truly in use by the job. Diane

[David Abramowitz]

How is the file being opened (input, update, etc.)? Dave