How to reduce the cost for this select script

Question

Hi,
"select id from order where cmd like 'Error ID %'"
This s my script. And the order table has morethan *5lakhs* records, when i select id for the above condition, my sql script cost is - *2200*,and cpu_cost is -- *235660697* i got this above information thro' plan_table, could anyone plz give suggestion to reduce the cost. (fyi -- cmd field has index with not null constraint).
Thanks.

Answer 1

user13294228 wrote:
"select id from order where cmd like 'Error ID %'"
This s my script. And the order table has morethan *5lakhs* records, when i select id for the above condition, my sql script cost is - *2200*,and cpu_cost is -- *235660697* i got this above information thro' plan_table, could anyone plz give suggestion to reduce the cost. (fyi -- cmd field has index with not null constraint).What does these cost numbers mean? Do you know what it means and what measurement unit is used for these costs? Will a cost of 2000 be fine? 1500? What specific number will make you sit back and think that the cost is now acceptable?
Or instead, do you not think it is more important to rather look at what the query does and why.. and then determine if there are methods to do that better and faster?
So let's look at what the query does. It uses a a LIKE predicate. This means that if an index is available, it cannot be used to find a specific indexed value... as the query is not sure what the value is. All that the query knows is what the first couple of characters are for the value.
So how would an index be used? Have you looked at the execution plan? Do you understand why the CBO made the decision it did?
Now - how do you expect the CBO to find the relevant rows any faster? The index benefits the query how? As the CBO cannot put the index to any better use than what it is already doing, what other options are there? Can alternative indexing or data structures be considered? Can parallel processing be used?
These questions are intended to make you analyse the problem - and understand the problem. That is always the 1st step.. solving the problem only comes after this 1st step.