MySQL query takes long to execute if using ORDER BY String Column - mysql

Question

So my query on a table that contains 4 million records executes instant if I dont use order by. However I want to give my clients a way to sort results by Name field and only show last 100 of the filtered result. As soon as I add order by Name it takes 100 seconds to execute.
My table structure is similar to this:
CREATE TABLE Test(
ID INT PRIMARY KEY AUTO_INCREMENT,
Name VARCHAR(100),
StatusID INT,
KEY (StatusID), <-- Index on StatusID
KEY (StatusID, Name) <-- Index on StatusID, Name
KEY(Name) <-- Index on Name
);
My query simply does something like:
explain SELECT ID, StatusID, Name
FROM Test
WHERE StatusID = 113
ORDER BY Name DESC
LIMIT 0, 100
Above explain when I order by Name gives this result:
StatusID_2 is the composite index of StatausID, Name
Now If I change ORDER BY Name DESC to ORDER BY ID I get this:
How can I make it so that it also examines only 100 rows when using ORDER BY Name?

Answer 1

You can try one thing, try for letters which would be in 100 rows expected in result like
SELECT *
FROM Test
*** Some Joins to filter data or get more columns from other tables
WHERE StatusID = 12 AND NAME REGEXP '^[A-H]'
ORDER BY Name DESC
LIMIT 0, 100
Moreover using index is very important on name (which is already applied) – in this case index range scan will be started and query execution stopped as soon as soon as required amount of rows generated.
So we can't use ID for nothing as it won't scan when it has reached its limit, the only thing we can try for is remove letters which are not possible in expected result and this what we are trying to do with REGEXP

Answer 2

It's hard to tell without the joins and the explain result but you're not making use of index appeareantly.
It might be because of the joins or because you have another key in the where clause. I'd recommend reading this, it covers all possible cases: http://dev.mysql.com/doc/refman/5.7/en/order-by-optimization.html
Increasing the sort_buffer_size and/or read_rnd_buffer_size might help...

Answer 3

You need a composite key based on the filtering WHERE criteria PLUS the order by... create an index on
( StatusID, Name )
This way, the WHERE jumps right to your StatusID = 12 records and ignores the rest of the 4 million... THEN uses the name as a secondary to qualify the ORDER BY.
Without seeing the other tables / join criteria and associated indexes, you might also want to try adding MySQL keyword
SELECT STRAIGHT_JOIN ... rest of query
So it does the query in the order you have selected, but unsure of impact without seeing other joins as noted previously.
ADDITION (per feedback)
I would remove the individual indexes on the ID only so the engine doesn't have to guess which one to use. The composite index can be used as an ID only query regardless of the name so you don't need to have both.
Also, remove the Name only index UNLESS you will ever be querying PRIMARILY on the name as a where qualifier without the ID qualifier... Also, how many total records are even possible for the example IDs you are querying out of the 4 million... You MIGHT want to pull the full set for the id as a sub-query, get a few thousand and have THAT ordered by name which would be quick... something like...
select *
from ( SELECT
ID,
StatusID,
Name
FROM
Test
WHERE
StatusID = 113 ) PreQuery
ORDER BY
Name DESC
LIMIT 0, 100