Comments on: How to produce random rows from a table

By: Amit S

Amit S — Sun, 31 Jul 2011 16:19:59 +0000

Interesting solution. I was wondering my this algorithm is not the native implementation in the ORDER BY RAND() function in MySql.

By: sagi baron

sagi baron — Tue, 15 Feb 2011 12:05:34 +0000

You can get a very efficient way to produce random rows from MySQL table by using the RAND function in the WHERE clause and not in the ORDER BY clause. It is returning random rows very fast and there is no need to worry about holes in the table's ids: See how to select random rows in mysql

By: RL

RL — Sun, 06 Feb 2011 04:39:45 +0000

I assume you used MyISAM tables for this solution? MyISAM stores the row counts in meta-information for each table… using COUNT(*) on other database engines may be a lot slower.

By: Michael

Michael — Fri, 14 Jan 2011 15:08:45 +0000

The workaround for the ORDER BY RAND()
its to add in where clause Rand

Example:
select * from mytbl where RAND()>0.9 ORDER BY RAND()

By: Justin

Justin — Tue, 16 Nov 2010 05:51:12 +0000

I’ve another solution, the problem I had with the solutions presented was extremely large and ramdom gaps in my index field, and the need to select not just one random row, but 50, and the need to use a where clause as well (selecting random records matching something else or other in the table). order by rand works but gets very slow when the table is huge and requires a sort.

My solution is to create a temporary table, which is effectively an view of the key field of the target table with a primary key that is auto_indexed. You can create this table in one like “create temporary view_table (id int auto_increment primary key, i) as select (i) from big_table where …”

then I create a list of random numbers between 0 and the number of rows in that temporary table and then do a join

select blah from big_table,view_table where view_table.id IN ($list) and view_table.i = big_table.i

and that’s it. Very fast.

Now and again (daily, hourly, whatever) drop the temporary table and re-create it.

By: Bruce

Bruce — Fri, 21 May 2010 03:26:18 +0000

Wow… interesting information.

By: Hazan Ilan

Hazan Ilan — Mon, 10 May 2010 08:09:03 +0000

Hi bdwankhede,
1. As far as I know (correct me if I’m wrong), there is no “FIRST” aggregate function in MySql, so you need to check the performance of it.
2. Please send me the Explain.
3. If “inspections” table is a big one, the ORDER BY RAND() is known to be a performance consumer. Please tell how much rows you have.
4. You can try selecting 1000 random records from “inspections” first and LEFT JOIN only on the results.

Waiting for your response
Ilan.

By: bdwankhede

bdwankhede — Tue, 04 May 2010 03:28:33 +0000

i have a query on Firebird sql as

SELECT FIRST 1000 i.”id”, i.”slug”, i.”what_they_do”, i.”does_well”, i.”last_improves” FROM “inspections” i LEFT JOIN “services” s on i.”service_id” = s.”id” WHERE s.”active” = 1 AND s.”suspended” = 0 AND i.”report_date” > ’2005-04-01′ ORDER BY RAND() ASC;

and it executes very slow , Please help…

By: Amsterdam Escort

Amsterdam Escort — Thu, 29 Apr 2010 21:44:43 +0000

I enjoyed viewing your blog and I will be back to check it more in the future so please keep up your good quality work. I love the colors that you chose, you are quite talented!

By: Bernarda Ocon

Bernarda Ocon — Thu, 29 Apr 2010 07:30:36 +0000

Wow… interesting information.