如何优化 MySQL 的 ORDERBYRAND()函数？

我想优化我的查询，所以我查看 mysql-slow.log。

我的大多数慢速查询包含 ORDER BY RAND()。我找不到解决这个问题的真正办法。在 MySQLPerformanceBlog有一个可能的解决方案，但我认为这还不够。在优化不佳(或经常更新，用户管理)的表上，它不起作用，或者在选择 PHP生成的随机行之前，我需要运行两个或更多查询。

这个问题有解决办法吗？

一个虚拟的例子:

SELECT  accomodation.ac_id,
accomodation.ac_status,
accomodation.ac_name,
accomodation.ac_status,
accomodation.ac_images
FROM    accomodation, accomodation_category
WHERE   accomodation.ac_status != 'draft'
AND accomodation.ac_category = accomodation_category.acat_id
AND accomodation_category.acat_slug != 'vendeglatohely'
AND ac_images != 'b:0;'
ORDER BY
RAND()
LIMIT 1
69327 次浏览
小开
最佳答案

Try this:

SELECT  *
FROM    (
SELECT  @cnt := COUNT(*) + 1,
@lim := 10
FROM    t_random
) vars
STRAIGHT_JOIN
(
SELECT  r.*,
@lim := @lim - 1
FROM    t_random r
WHERE   (@cnt := @cnt - 1)
AND RAND(20090301) < @lim / @cnt
) i

This is especially efficient on MyISAM (since the COUNT(*) is instant), but even in InnoDB it's 10 times more efficient than ORDER BY RAND().

The main idea here is that we don't sort, but instead keep two variables and calculate the running probability of a row to be selected on the current step.

See this article in my blog for more detail:

Update:

If you need to select but a single random record, try this:

SELECT  aco.*
FROM    (
SELECT  minid + FLOOR((maxid - minid) * RAND()) AS randid
FROM    (
SELECT  MAX(ac_id) AS maxid, MIN(ac_id) AS minid
FROM    accomodation
) q
) q2
JOIN    accomodation aco
ON      aco.ac_id =
COALESCE
(
(
SELECT  accomodation.ac_id
FROM    accomodation
WHERE   ac_id > randid
AND ac_status != 'draft'
AND ac_images != 'b:0;'
AND NOT EXISTS
(
SELECT  NULL
FROM    accomodation_category
WHERE   acat_id = ac_category
AND acat_slug = 'vendeglatohely'
)
ORDER BY
ac_id
LIMIT   1
),
(
SELECT  accomodation.ac_id
FROM    accomodation
WHERE   ac_status != 'draft'
AND ac_images != 'b:0;'
AND NOT EXISTS
(
SELECT  NULL
FROM    accomodation_category
WHERE   acat_id = ac_category
AND acat_slug = 'vendeglatohely'
)
ORDER BY
ac_id
LIMIT   1
)
)

This assumes your ac_id's are distributed more or less evenly.

小开

Here's how I'd do it:

SET @r := (SELECT ROUND(RAND() * (SELECT COUNT(*)
FROM    accomodation a
JOIN    accomodation_category c
ON (a.ac_category = c.acat_id)
WHERE   a.ac_status != 'draft'
AND c.acat_slug != 'vendeglatohely'
AND a.ac_images != 'b:0;';


SET @sql := CONCAT('
SELECT  a.ac_id,
a.ac_status,
a.ac_name,
a.ac_status,
a.ac_images
FROM    accomodation a
JOIN    accomodation_category c
ON (a.ac_category = c.acat_id)
WHERE   a.ac_status != ''draft''
AND c.acat_slug != ''vendeglatohely''
AND a.ac_images != ''b:0;''
LIMIT ', @r, ', 1');


PREPARE stmt1 FROM @sql;


EXECUTE stmt1;
小开

It depends on how random you need to be. The solution you linked works pretty well IMO. Unless you have large gaps in the ID field, it's still pretty random.

However, you should be able to do it in one query using this (for selecting a single value):

SELECT [fields] FROM [table] WHERE id >= FLOOR(RAND()*MAX(id)) LIMIT 1

Other solutions:

  • Add a permanent float field called random to the table and fill it with random numbers. You can then generate a random number in PHP and do "SELECT ... WHERE rnd > $random"
  • Grab the entire list of IDs and cache them in a text file. Read the file and pick a random ID from it.
  • Cache the results of the query as HTML and keep it for a few hours.
小开

This will give you single sub query that will use the index to get a random id then the other query will fire getting your joined table.

SELECT  accomodation.ac_id,
accomodation.ac_status,
accomodation.ac_name,
accomodation.ac_status,
accomodation.ac_images
FROM    accomodation, accomodation_category
WHERE   accomodation.ac_status != 'draft'
AND accomodation.ac_category = accomodation_category.acat_id
AND accomodation_category.acat_slug != 'vendeglatohely'
AND ac_images != 'b:0;'
AND accomodation.ac_id IS IN (
SELECT accomodation.ac_id FROM accomodation ORDER BY RAND() LIMIT 1
)
小开

The solution for your dummy-example would be:

SELECT  accomodation.ac_id,
accomodation.ac_status,
accomodation.ac_name,
accomodation.ac_status,
accomodation.ac_images
FROM    accomodation,
JOIN
accomodation_category
ON accomodation.ac_category = accomodation_category.acat_id
JOIN
(
SELECT CEIL(RAND()*(SELECT MAX(ac_id) FROM accomodation)) AS ac_id
) AS Choices
USING (ac_id)
WHERE   accomodation.ac_id >= Choices.ac_id
AND accomodation.ac_status != 'draft'
AND accomodation_category.acat_slug != 'vendeglatohely'
AND ac_images != 'b:0;'
LIMIT 1

To read more about alternatives to ORDER BY RAND(), you should read this article.

小开

I am optimizing a lot of existing queries in my project. Quassnoi's solution has helped me speed up the queries a lot! However, I find it hard to incorporate the said solution in all queries, especially for complicated queries involving many subqueries on multiple large tables.

So I am using a less optimized solution. Fundamentally it works the same way as Quassnoi's solution.

SELECT  accomodation.ac_id,
accomodation.ac_status,
accomodation.ac_name,
accomodation.ac_status,
accomodation.ac_images
FROM    accomodation, accomodation_category
WHERE   accomodation.ac_status != 'draft'
AND accomodation.ac_category = accomodation_category.acat_id
AND accomodation_category.acat_slug != 'vendeglatohely'
AND ac_images != 'b:0;'
AND rand() <= $size * $factor / [accomodation_table_row_count]
LIMIT $size

$size * $factor / [accomodation_table_row_count] works out the probability of picking a random row. The rand() will generate a random number. The row will be selected if rand() is smaller or equals to the probability. This effectively performs a random selection to limit the table size. Since there is a chance it will return less than the defined limit count, we need to increase probability to ensure we are selecting enough rows. Hence we multiply $size by a $factor (I usually set $factor = 2, works in most cases). Finally we do the limit $size

The problem now is working out the accomodation_table_row_count. If we know the table size, we COULD hard code the table size. This would run the fastest, but obviously this is not ideal. If you are using Myisam, getting table count is very efficient. Since I am using innodb, I am just doing a simple count+selection. In your case, it would look like this:

SELECT  accomodation.ac_id,
accomodation.ac_status,
accomodation.ac_name,
accomodation.ac_status,
accomodation.ac_images
FROM    accomodation, accomodation_category
WHERE   accomodation.ac_status != 'draft'
AND accomodation.ac_category = accomodation_category.acat_id
AND accomodation_category.acat_slug != 'vendeglatohely'
AND ac_images != 'b:0;'
AND rand() <= $size * $factor / (select (SELECT count(*) FROM `accomodation`) * (SELECT count(*) FROM `accomodation_category`))
LIMIT $size

The tricky part is working out the right probability. As you can see the following code actually only calculates the rough temp table size (In fact, too rough!): (select (SELECT count(*) FROM accomodation) * (SELECT count(*) FROM accomodation_category)) But you can refine this logic to give a closer table size approximation. Note that it is better to OVER-select than to under-select rows. i.e. if the probability is set too low, you risk not selecting enough rows.

This solution runs slower than Quassnoi's solution since we need to recalculate the table size. However, I find this coding a lot more manageable. This is a trade off between accuracy + performance vs coding complexity. Having said that, on large tables this is still by far faster than Order by Rand().

Note: If the query logic permits, perform the random selection as early as possible before any join operations.

小开

(Yeah, I will get dinged for not having enough meat here, but can't you be a vegan for one day?)

Case: Consecutive AUTO_INCREMENT without gaps, 1 row returned
Case: Consecutive AUTO_INCREMENT without gaps, 10 rows
Case: AUTO_INCREMENT with gaps, 1 row returned
Case: Extra FLOAT column for randomizing
Case: UUID or MD5 column

Those 5 cases can be made very efficient for large tables. See my blog for the details.

小开 
function getRandomRow(){
$id = rand(0,NUM_OF_ROWS_OR_CLOSE_TO_IT);
$res = getRowById($id);
if(!empty($res))
return $res;
return getRandomRow();
}


//rowid is a key on table
function getRowById($rowid=false){


return db select from table where rowid = $rowid;
}
小开

My recommendation is to add a column with a UUID (version 4) or other random value, with a unique index (or just the primary key).

Then you can simply generate a random value at query time and select rows greater than the generated value, ordering by the random column.

Make sure if you receive less than the expected number of rows, you repeat the query without the greater than clause (to select rows at the "beginning" of the result set).

uuid = generateUUIDV4()


select * from foo
where uuid > :uuid
order by uuid
limit 42


if count(results) < 42 {
select * from foo
order by uuid
limit :remainingResultsRequired
}

提交答案