Re: Performance Problem - how to tune a query?

[HubbleSpace]

How many entries are in this table? Do you use an index?

I think if you use an index on data_starttime (I assume data_starttime is a field of billing_data ) and change you request like
...
WHERE data_starttime IN BETWEEN '20051001' AND '20051031'
AND MID(...
it would return faster.

The statement LIKE need to do an full table scan(in my opinion) . So if you got more than ??100000(or much more) lines in billing_dat that'll take a while.

It depends too on the number data that is filtered by the data_starttime. If its more that 1% of the total data an index does not work so much faster.

Another thing is the MID(d.data_site,3,3) function. It is called for each row.
So if many rows are returned it slows down your request too.

Hubble