Slow sql execution

Question

I have SQL query retreiving lots of data from several tables, and having specific condition for nohem as it is displayed in following subquery:
ss.nohem=(select Max(ss.nohem)
from sschem ss
where nohem in (3,5,6)
and TRIM(ss.id)=TRIM(hc.id)
As I notice, this subquery slows down main SQL execution (besides the tables are large and not well organazied).
select ss.nohem, hc.time, hc.id, hc.shift, hc.crew, tm.qualityorg,
ss.c, ss.si, ss.mn, ..., ..., ss.ca,
l2_min.c as c_min, l2_max.c as c_max,
l2_min.si as si_min, l2_max.si as si_max,
l2_min.mn as mn_min, l2_max.mn as mn_max,
l2_min.ca as ca_min, l2_max.ca as ca_max,
tm.quality
from sschem ss, tmelt tm, h_cast hc,
l2_chem_grade l2_min, l2_chem_grade l2_max
where (hc.time > {?from} and hc.time <= {?to})
and ss.nohem=(select Max(ss.nohem)
from sschem ss
where nohem in (3,5,6)
and TRIM(ss.id)=TRIM(hc.id)
and tm.qualityorg=l2_min.quality_code and l2_min.range_type='MIN'
and tm.qualityorg=l2ch_max.quality_code and l2_max.range_type='MAX'
and TRIM(ss.id)=TRIM(tm.id)
and TRIM(hc.id)=TRIM(tm.id)
and TRIM(ss.id)=TRIM(hc.id)
order by hc.time
How can I reorganize this query to make its execution faster?
Thanks!

Answer 1

Problem with the sub-query is likely a full table scan needs to be performed. The predicate "TRIM(ss.id)=TRIM(hc.id)" uses a function to convert the searched column.
This means reading every row applicable, applying the TRIM function to the ID column, and then evaluating the predicate. If the ID column is indexed, the index is made useless by the function.
Solution in this case - either make use of a function index, or better still, apply the TRIM via the INSERT/UPDATE trigger and create a normal index on the ID column.
Second problem is that the sub-select is ran for every single row selected in the main loop - it is a corellated query as the predicate for the sub-query includes a value from the row from the main query.
So if the main query hits 10,000 rows, this sub-query will be executed 10,000 times. With the sub-query doing a full table scan, this is very, very expensive.
Solutions. Ensure that all filters and processing are done (upfront) to reduce the result of the main query to a minimum, in order to reduce the number of sub-queries to perform (1 per row) to a minimum. Or rethink the logic and design of the entire query and do away together with the corellated sub-query.