MYSQL 查询慢 SELECT DISTINCT

Question

I want to output the most used search items of my classifieds of the last 3 days

I have an extra table where I save all search items together with the day and a counter which increments if the search is used again (and for future use the category page it was used on)

This works but the speed is less than desireable.

Any Ideas on how to improve the performance?

UPDATE: It seems that copying to temp table is causing the problem- How can i change this?

SQL looks like:

SELECT distinct search_item 
FROM `tb_search` 
WHERE added >= ( CURDATE() - INTERVAL 3 DAY ) 
ORDER BY count DESC 
LIMIT 0,20

EXPLAIN:

id select_type table        type    possible_keys  key     key_len ref rows Extra
1   SIMPLE     tb_search    range   added          added    3       NULL 4537   Using index condition; Using temporary; Using filesort...   

But it is quite slow, it takes 12 ms of the complete 40ms execution time of the php script so 25%

The table structure (Innodb) 23.000 records

tb_search
Column  Type    Null    Default Links to    Comments    Media (MIME) type
id (Primary)    bigint(20)  No              
search_item varchar(30) No              
category    varchar(30) Yes NULL            
added   date    No              
count   smallint(6) No              


 Indexes
    Keyname Type    Unique  Packed  Column  Cardinality Collation   Null    Comment
    PRIMARY BTREE   Yes No  id  23744   A   No  
    search_item BTREE   No  No  search_item 11872   A   No  
    added   BTREE   No  No  added   44  A   No  

space usage

Data 1.5 MiB Index 1.9 MiB Overhead 466.0 MiB <-- This is the same for all tables??? optimizing all tables did not help? Effective -485,048,320 B Total 3.4 MiB

Is the overhead a problem? How can I get rid of it? Optimizing all tables did not change it. Total size of db is 86MB

**Detailed profile**

1   Starting    48 µs
2   Checking Permissions    6 µs
3   Opening Tables  14 µs
4   After Opening Tables    13 µs
5   System Lock 5 µs
6   Table Lock  3 µs
7   After Table Lock    5 µs
8   Init    20 µs
9   Optimizing  11 µs
10  Statistics  85 µs
11  Preparing   34 µs
12  Executing   4 µs
13  Creating Tmp Table  24 µs
**14    Copying To Tmp Table    9.7 ms**
15  Sorting Result  218 µs
16  Sending Data    18 µs
17  End 4 µs
18  Removing Tmp Table  9 µs
19  End 5 µs
20  Query End   7 µs
21  Closing Tables  9 µs
22  Freeing Items   8 µs
23  Updating Status 21 µs
24  Cleaning Up 4 µs
Summary by stateDocumentation
State
Total Time
% Time
Calls
ø Time
Starting    48 µs   0.47%   1   48 µs
Checking Permissions    6 µs    0.06%   1   6 µs
Opening Tables  14 µs   0.14%   1   14 µs
After Opening Tables    13 µs   0.13%   1   13 µs
System Lock 5 µs    0.05%   1   5 µs
Table Lock  3 µs    0.03%   1   3 µs
After Table Lock    5 µs    0.05%   1   5 µs
Init    20 µs   0.19%   1   20 µs
Optimizing  11 µs   0.11%   1   11 µs
Statistics  85 µs   0.83%   1   85 µs
Preparing   34 µs   0.33%   1   34 µs
Executing   4 µs    0.04%   1   4 µs
Creating Tmp Table  24 µs   0.23%   1   24 µs
**Copying To Tmp Table  9.7 ms  94.42%  1   9.7 ms**
Sorting Result  218 µs  2.12%   1   218 µs
Sending Data    18 µs   0.17%   1   18 µs
End 4 µs    0.04%   2   2 µs
Removing Tmp Table  9 µs    0.09%   1   9 µs
Query End   7 µs    0.07%   1   7 µs
Closing Tables  9 µs    0.09%   1   9 µs
Freeing Items   8 µs    0.08%   1   8 µs
Updating Status 21 µs   0.20%   1   21 µs
Cleaning Up 4 µs    0.04%   1   4 µs

table structure

-- phpMyAdmin SQL Dump
-- version 5.0.3
-- https://www.phpmyadmin.net/
--
-- Host: localhost:3306
-- Generation Time: Apr 10, 2021 at 11:03 AM
-- Server version: 5.5.68-MariaDB
-- PHP Version: 7.3.27

SET SQL_MODE = "NO_AUTO_VALUE_ON_ZERO";
START TRANSACTION;
SET time_zone = "+00:00";


/*!40101 SET @OLD_CHARACTER_SET_CLIENT=@@CHARACTER_SET_CLIENT */;
/*!40101 SET @OLD_CHARACTER_SET_RESULTS=@@CHARACTER_SET_RESULTS */;
/*!40101 SET @OLD_COLLATION_CONNECTION=@@COLLATION_CONNECTION */;
/*!40101 SET NAMES utf8mb4 */;

--
-- Database: `terraristik_main`
--

-- --------------------------------------------------------

--
-- Table structure for table `tb_search`
--

CREATE TABLE `tb_search` (
  `id` bigint(20) UNSIGNED NOT NULL,
  `search_item` varchar(30) NOT NULL,
  `category` varchar(30) DEFAULT NULL,
  `added` date NOT NULL,
  `count` smallint(6) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

--
-- Indexes for dumped tables
--

--
-- Indexes for table `tb_search`
--
ALTER TABLE `tb_search`
  ADD PRIMARY KEY (`id`),
  ADD KEY `search_item` (`search_item`),
  ADD KEY `added` (`added`);

--
-- AUTO_INCREMENT for dumped tables
--

--
-- AUTO_INCREMENT for table `tb_search`
--
ALTER TABLE `tb_search`
  MODIFY `id` bigint(20) UNSIGNED NOT NULL AUTO_INCREMENT;
COMMIT;

/*!40101 SET CHARACTER_SET_CLIENT=@OLD_CHARACTER_SET_CLIENT */;
/*!40101 SET CHARACTER_SET_RESULTS=@OLD_CHARACTER_SET_RESULTS */;
/*!40101 SET COLLATION_CONNECTION=@OLD_COLLATION_CONNECTION */;

Answer 1

SOLUTION

By adding an index for count I reduced the time by more than 95%..! Now the whole operation takes abou 1-1.5 ms.

The copy_to_temp table is down from 10 ms to 0.5 ms - still using 65% of all time but sufficently fast for my needs.

Answer 2

Use for a date three days ago:

SELECT distinct search_item 
FROM `tb_search t` 
WHERE t.date >= DATE_ADD(CURDATE(), INTERVAL -3 DAY);

Or you can use:

SELECT distinct search_item 
FROM `tb_search t` 
WHERE t.date >= ( CURDATE() - INTERVAL 3 DAY );