Regards,
Sarfaraz Ahmed
Microsoft Excel help

This will help me to research further.

Any time you apply a function to a column in the where/join clause, you eliminate the ability of the query optimizer to use indexes on the column. Further, if the query optimizer doesn’t have other good indexes to use, it may “invert” the query (as I call it) where it uses the primary key of the table to do a cartesian product join with another table, and then apply the functions to the resulting giant intermediate result set.
Avoid functions on columns in a where/join clause like the plague…

Some of your joins are using “ltrim(rtrim(<column>))”. Much better would be to enforce removal of leading/trailing blanks on data entry (with a trigger, perhaps), which would allow you to do direct compares of the data in the columns without the functions, and therefore allow use of indexes on those columns.

Also note that “len(ltrim(remarks)) > 0″ should be the same as “remarks is not null” if you eliminate leading/trailing blanks on entry. The “len(trim())” expression could cause a table scan, while “remarks is not null” *could* use an index on that column (it might not, depending on the actual distribution of data in the records).

Finally, if there is no good way to avoid leading/trailing blanks in the data, you might consider creating materialized views of the data where the view contains the blank-stripped data and has a column containing the field lengths where needed. Then, your query can go against that view. Depending on the volume of updates to the table and view, this could dramatically speed up the query. (Note that a “materialized” view is a “real” table containing the massaged data from the underlying tables, and that this table can be indexed like any other tables – just creating a standard “view” will NOT generally gain any performance, as the view will be constructed at query time with no advantage over just running the query.)

SQL Server Query Execution Plan Analysis
Execution Plan Basics
Capture execution plans with SQL Server 2005 Profiler

Thanks for your valuable answer.

I can see this kind of effect in any simple sql query.

Can you point me to the right source where I can find detailed information on how to read the execution plan and do the performance tuning.

Thanks,
John