I have a particular query in MySQL where it compares values from two different table (indexed) using collate but this does not execute for hours. The query is provided below:
create temporary table elig_temp
select id from table_elig;
create index elig_temp on elig_temp(id);
create temporary table med_temp
select id from table_med;
create index med_temp on med_temp(id);
select COUNT(1) as result
from med_temp a
where exists
(
select 1
from elig_temp b
where a.id collate latin1_general_cs = b.id collate latin1_general_cs
)
FYI The elig_temp table has 70k records whereas the med_temp has 1 million records.
Also, the id fields of table_elig and table_med tables are the hash encrypted values of another field from the same table. Therefore, I tried to use binary collations techniques too, such as udf8_bin and latin1_bin to make the query run but I am stuck again.
I have even tried by defining with the same collation techniques, that I used with query, for each fields (varchar and char) of table_med and table_elig but no luck.
Please suggest me with any possible solution for executing this query time efficiently.
When you explicitly set a collate, then MySQL cannot use indexes on the column. This is unfortunate.
First, does the query run without collations?
select COUNT(1) as result
from med_temp a
where exists (select 1 from elig_temp b where a.id = b.id collate );
That is the simplest solution.
The next solution is to make the collations the same when you create the tables:
create temporary table elig_temp
select id collate latin1_general_cs as id
from table_elig;
create index elig_temp on elig_temp(id);
create temporary table med_temp
select id collate latin1_general_cs as id
from table_med;
create index med_temp on med_temp(id);
Then you can run the above query -- without the explicit collations -- and it should use the indexes.