Question

I would like to know what are the best solutions to optimize the response time of an SQL query for a table containing more than 2.000.000 records

As a solution, I thought of a virtual table by creating a SQL view (In fact, I prefer mysql to search early in the lines created this year because the data of this application is based on the season.)

Is there a better solution or recommendation?

eg to search all rent lines of rent 12

before =>
select * from rent_lines Where rent_id = 12

Now =>
  I created a view

CREATE VIEW v_rent_lines
AS SELECT rent_id, category_id, customer_id, amount ..
Where rent_lines FROM created_at > = (select starts_on from seasons where current = true)

select * from v_rent_lines Where rent_id = 12

Notes:

• database engine is being used InnoDB

• I added indexes table (index_rent_lines_on_rent_id, index_rent_lines_on_category_id, index_rent_lines_on_customer_id)

• rent has many rent_lines

Answer 1

The view is not really helping anything in this case. Views primarily help you, as a developer, by letting you refer to a more complex query by name rather than having to repeat the details all the time. They don’t really help mysql all that much.

You have a lot of options.

1. If you don’t yet, ensure that you have an index where rent_id is either the only field in the index, or the first field in the index. For example:

   create index rent_id_idx on rent_lines (rent_id)
   

2. You can consider using mysql’s partitioning system and partitioning on rent_id

3. You can create an index that will have a lower cardinality by doing something like:

   alter table rent_lines add rent_id_bucket smallint unsigned not null after rent_id;
   update rent_lines set rent_id_bucket = rent_id>>16;
   alter table rent_lines add key rent_id_bucket_idx(rent_id_bucket);
   

   This will let you do a query like:

   select * from rent_lines where rent_id_bucket = 16>>16 and rent_id=16