Multiple simple joins Vs. Multiple simple selects - mysql

I have a meta table that looks like this
ID post_id meta_key meta_value
1 1 key_01 val
2 1 key_02 val
3 1 key_03 val
and I want to get all 3 keys and values. Is it any slower to use JOIN then SELECT? JOIN would make it so much easier, for example:
SELECT M1.meta_value AS key_01, M2.meta_value AS key_02, M3.meta_value AS key_04
FROM `meta` AS M1
JOIN `meta` M2 ON M2.post_id = 1 AND M2.meta_key = 'key_02'
JOIN `meta M3 ON M3.post_id = 1 AND M3.meta_key = 'key_03'
WHERE M1.meta_key = 'key_01'
I didn't test that any, but you should get the idea.

I beleive JOINS are more efficient than using nested SELECTS. Also indexing the foreign-key columns (joined columns) will improve performance as well. So pop an index on post_id and meta_key. Indexes may hurt INSERT performance so take this into consideration.

i think that the mysql engine has a better chance to create a good query plan when using joins than it has when doing sub-selects and/or unions.

Related

Query optimization (multiple joins)

I would like to find a way to improve a query but it seems i've done it all. Let me give you some details.
Below is my query :
SELECT
`u`.`id` AS `id`,
`p`.`lastname` AS `lastname`,
`p`.`firstname` AS `firstname`,
COALESCE(`r`.`value`, 0) AS `rvalue`,
SUM(`rat`.`category` = 'A') AS `count_a`,
SUM(`rat`.`category` = 'B') AS `count_b`,
SUM(`rat`.`category` = 'C') AS `count_c`
FROM
`user` `u`
JOIN `user_customer` `uc` ON (`u`.`id` = `uc`.`user_id`)
JOIN `profile` `p` ON (`p`.`id` = `u`.`profile_id`)
JOIN `ad` FORCE INDEX (fk_ad_customer_idx) ON (`uc`.`customer_id` = `ad`.`customer_id`)
JOIN `ac` ON (`ac`.`id` = `ad`.`ac_id`)
JOIN `a` ON (`a`.`id` = `ac`.`a_id`)
JOIN `rat` ON (`rat`.`code` = `a`.`rat_code`)
LEFT JOIN `r` ON (`r`.`id` = `u`.`r_id`)
GROUP BY `u`.`id`
;
Note : Some table and column names are voluntarily hidden.
Now let me give you some volumetric data :
user => 6534 rows
user_customer => 12 923 rows
profile => 6511 rows
ad => 320 868 rows
ac => 4505 rows
a => 536 rows
rat => 6 rows
r => 3400 rows
And finally, my execution plan :
My query does currently run in around 1.3 to 1.7 seconds which is slow enough to annoy users of my application of course ... Also fyi result set is composed of 165 rows.
Is there a way I can improve this ?
Thanks.
EDIT 1 (answer to Rick James below) :
What are the speed and EXPLAIN when you don't use FORCE INDEX?
Surprisingly it gets faster when i don't use FORCE INDEX. To be honest, i don't really remember why i've done that change. I've probably found better results in terms of performance with it during one of my various tries and didn't remove it since.
When i don't use FORCE INDEX, it uses an other index ad_customer_ac_id_blocked_idx(customer_id, ac_id, blocked) and times are around 1.1 sec.
I don't really get it because fk_ad_customer_idx(customer_id) is the same when we talk about index on customer_id.
Get rid of FORCE INDEX. Even if it helped yesterday; it may hurt tomorrow.
Some of these indexes may be beneficial. (It is hard to predict; so simply add them all.)
a: (rat_code, id)
rat: (code, category)
ac: (a_id, id)
ad: (ac_id, customer_id)
ad: (customer_id, ac_id)
uc: (customer_id, user_id)
uc: (user_id, customer_id)
u: (profile_id, r_id, id)
(This assumes that id is the PRIMARY KEY of each table. Note that none have id first.) Most of the above are "covering".
Another approach that sometimes helps: Gather the SUMs before joining to any unnecessary table. But is seems that p is the only table not involved in getting from u (the target of GROUP BY) to r and rat (used in aggregates). It would look something like:
SELECT ..., firstname, lastname
FROM ( everything as above except for `p` ) AS most
JOIN `profile` `p` ON (`p`.`id` = most.`profile_id`)
GROUP BY most.id
This avoids hauling around firstname and lastname while doing most of the joins and the GROUP BY.
When doing JOINs and GROUP BY, be sure to sanity check the aggregates. Your COUNTs and SUMs may be larger than they should be.
First, you don't need to tick.everyTableAndColumn in your queries, nor result columns, aliases, etc. The tick marks are used primarily when you are in conflict with a reserved work so the parser knows you are referring to a specific column... like having a table with a COLUMN named "JOIN", but JOIN is part of SQL command... see the confusion it would cause. Helps clean readability too.
Next, and this is just personal preference and can help you and others following behind you on data and their relationships. I show the join as indented from where it is coming from. As you can see below, I see the chain on how do I get from the User (u alias) to the rat alias table... You get there only by going 5 levels deep, and I put the first table on the left-side of the join (coming from table) then = the table joining TO right-side of join.
Now, that I can see the relationships, I would suggest the following. Make COVERING indexes on your tables that have the criteria, and id/value where appropriate. This way the query gets as best it needs, the data from the index page vs having to go to the raw data. So here are suggestions for indexes.
table index
user_customer ( user_id, customer_id ) -- dont know what your fk_ad_customer_idx parts are)
ad ( customer_id, ac_id )
ac ( id, a_id )
a (id, rat_code )
rat ( code, category )
Reformatted query for readability and seeing relationships between the tables
SELECT
u.id,
p.lastname,
p.firstname,
COALESCE(r.value, 0) AS rvalue,
SUM(rat.category = 'A') AS count_a,
SUM(rat.category = 'B') AS count_b,
SUM(rat.category = 'C') AS count_c
FROM
user u
JOIN user_customer uc
ON u.id = uc.user_id
JOIN ad FORCE INDEX (fk_ad_customer_idx)
ON uc.customer_id = ad.customer_id
JOIN ac
ON ad.ac_id = ac.id
JOIN a
ON ac.a_id = a.id
JOIN rat
ON a.rat_code = rat.code
JOIN profile p
ON u.profile_id = p.id
LEFT JOIN r
ON u.r_id = r.id
GROUP BY
u.id

very slow query with ORDER BY

I have read through many tutorials online and here on stackoverflow but I still can't figure out how to solve the problem I'm facing right now.
I would like to tell you guys that I'm a mysql newbie so please forgive my noobness.
Alright, the query is this and it grabs the information that I need from wordpress database
SELECT
product.ID productId,
product.guid productLink,
product.post_title productTitle,
post.ID postId,
post.post_title postTitle,
post.post_content postContent,
post.post_date postDate,
tm.slug typeSlug, tm.name typeName,
tm2.slug langSlug, tm2.name langName,
tm3.slug pubSlug, tm3.name pubName,
IFNULL(wl.id,0) wishlist
FROM wp_posts product
JOIN wp_postmeta meta ON meta.meta_key = 'p2m' AND meta.meta_value=product.ID
JOIN wp_posts post ON post.ID = meta.post_id
JOIN wp_term_relationships tr ON tr.object_id = product.ID
JOIN wp_term_taxonomy tt ON tt.term_taxonomy_id = tr.term_taxonomy_id AND tt.taxonomy = 'mtype'
JOIN wp_terms tm ON tm.term_id = tt.term_id
JOIN wp_term_relationships tr2 ON tr2.object_id = product.ID
JOIN wp_term_taxonomy tt2 ON tt2.term_taxonomy_id = tr2.term_taxonomy_id AND tt2.taxonomy = 'language'
JOIN wp_terms tm2 ON tm2.term_id = tt2.term_id
JOIN wp_term_relationships tr3 ON tr3.object_id = product.ID
JOIN wp_term_taxonomy tt3 ON tt3.term_taxonomy_id = tr3.term_taxonomy_id AND tt3.taxonomy = 'publisher'
JOIN wp_terms tm3 ON tm3.term_id = tt3.term_id
LEFT JOIN wp_yith_wcwl wl ON wl.user_id = 1 AND wl.prod_id = product.ID AND wl.post_id = post.ID
WHERE product.post_type = 'product'
ORDER BY post.post_date DESC LIMIT 0,35
When I remove "ORDER BY post.post_date DESC" the speed of the query gets down to .03 seconds which is freaking amazing.. But with the addition of the "ORDER BY post.post_date DESC" the speed of the query goes to amazing 10+ seconds which is way too long..
I've used EXPLAIN and it seems that there is usage of filesort when the ORDER BY by date gets into the query.
I need to have my query reply back the results according to the post_date so I can't figure out what I could do at this point...
Additionally, I would like to point it out that in Database Description of wordpress there is an INDEX referred as "type_status_date" which could be used in my case. However, I'm totally clueless where to use it and how to do it. If there is anyone who can point out the flaw in the logic of my query or help me out with the optimization of the query (or index) please do so. Thanks for you kind attention!
P.S: I don't know how to create an index too :)
Initial Result of EXPLAIN with ORDER BY
JOIN wp_postmeta meta
ON meta.meta_key = 'p2m' -- filters
AND meta.meta_value=product.ID -- shows relation
is confusing. JOIN...ON is used to say how two tables are related. Filters belong in WHERE:
WHERE ...
AND meta.meta_key = 'p2m'
...
wp_postmeta is not well indexed. More discussion here .
Adding INDEX(post_date) may or may not help performance -- It depends on how quickly 35 good rows are found.
From the EXPLAIN, we see that the worst part is getting into meta -- something like 30K rows to look through. This _estimates that there are 30 rows with meta_key = 'p2m'. How many rows are there?
Unfortunately wp_postmeta is not designed to efficiently start with the meta_key+meta_value. This is a general problem with key-value stores (such as Posts in WP), especially when the 'value' is LONGTEXT.
The index on wp_post for type_status_date, has the date field as the third field of the index,
type_status_date INDEX
post_type
post_status
post_date
ID
So you have a predicate for post type, but post status is not included in your query in any predicate, at best it can therefore do a partial index scan of the index (sometimes called a skip scan or range scan depending on the db, my sql is not going to play ball easily on that) but it will be slower.
That is a best case scenario though, with all those joins and additional fields not covered by the index the cost of the index scan and row lookups could be far too high to consider even touching the index vs a straight scan.
It would help if you post the explain plan, it would help confirm what the optimizer was doing. The above is a more generic DB engine commentary.
type_status_date is a combined index so it's used only if you order by all it's components. It cannot be used by MySQL to order by only post_date. So the best solution is to add an index by post_date.

Mysql - db-select with several joins and tablestructure

I am using the following mysql select-query with several joins. I am wondering if this is how a somewhat good select-statement should look like:
SELECT *
FROM table_news AS a
INNER JOIN table_cat AS b ON a.cat_id = b.id
INNER JOIN table_countries AS c ON a.country_id = c.id
INNER JOIN table_addresses AS d ON a.id = d.news_id
WHERE a.deleted = 0
AND a.hidden = 0
AND a.cat_id = ".$search_cat."
AND a.country_id = ".$search_country."
AND a.title LIKE '%".$search_string."%'
OR a.deleted = 0
AND a.hidden = 0
AND a.cat_id = ".$search_cat."
AND a.country_id = ".$search_country."
AND a.subtitle LIKE '%".$search_string."%'"
It seems to be a lot of joins. Even though table b and table c contain only 3 or 4 fields, I wonder if the number of joins would clearly slow down the search on the starting-page?
Would it be better to put the fields from table d (street, city and so on) back into the main-table, as they should be needed most of the time this query is executed?
Thanx in advance,
Jayden
I don't think there is necessarily anything wrong with having three joins. There are a couple of things you can do to make sure the query is optimised.
Firstly, you should never do SELECT * - instead explicitly state what fields you want to return from the database.
Also, I would create indexes on all the fields you have in the where clause, and all of the fields you are joining. This can be a little bit of a trade off - for example if you are doing a lot of write operations then there is a hit because you need to write to the index everytime.

Searching on multi (1-n) relation tables

I learned the hard way that i shouldn't store serialized data in a table when i need to make it searchable .
So i made 3 tables the base & two 1-n relation tables .
So here is the query i get if i want to select a specific activity .
SELECT
jdc_organizations_activities.id
FROM
jdc_activity_sector ,
jdc_activity_type
INNER JOIN jdc_organizations_activities ON jdc_activity_type.activityId = jdc_organizations_activities.id
AND
jdc_activity_sector.activityId = jdc_organizations_activities.id
WHERE
jdc_activity_sector.activitySector = 5 AND
jdc_activity_type.activityType = 3
Questions :
1- What kind of indexes can i add on a 1-n relation table , i already have a unique combination of (activityId - activitySector) & (activityId - activityType)
2- Is there a better way to write the query to have a better performance ?
Thank you !
I would re-organise the query to avoid the cross product caused by using , notation.
Also, you are effectively only using the sector and type tables as filters. So put activity table first, and then join on your other tables.
Some may suggest that; the first join should ideally be the join which is most likely to restrict your results the most, leaving the minimal amount of work to do in the second join. In reality, the sql engine can actually re-arrange your query when generateing a plan, but it does help to think this way to help you think about the efforts the sql engine are having to go to.
Finally, there are the indexes on each table. I would actually suggest reversing the Indexes...
- ActivitySector THEN ActivityId
- ActivityType THEN ActivityId
This is specifically because the sql engine is manipulating your query. It can take the WHERE clause and say "only include records from the Sector table where ActivitySector = 5", and similarly for the Type table. By having the Sector and Type identifies FIRST in the index, this filtering of the tables can be done much faster, and then the joins will have much less work to do.
SELECT
[activity].id
FROM
jdc_organizations_activities AS [activity]
INNER JOIN
jdc_activity_sector AS [sector]
ON [activity].id = [sector].activityId
INNER JOIN
jdc_activity_type AS [type]
ON [activity].id = [type].activityId
WHERE
[sector].activitySector = 5
AND [type].activityType = 3
Or, because you don't actually use the content of the Activity table...
SELECT
[sector].activityId
FROM
jdc_activity_sector AS [sector]
INNER JOIN
jdc_activity_type AS [type]
ON [sector].activityId = [type].activityId
WHERE
[sector].activitySector = 5
AND [type].activityType = 3
Or...
SELECT
[activity].id
FROM
jdc_organizations_activities AS [activity]
WHERE
EXISTS (SELECT * FROM jdc_activity_sector WHERE activityId = [activity].id AND activitySector = 5)
AND EXISTS (SELECT * FROM jdc_activity_type WHERE activityId = [activity].id AND activityType = 3)
I would advise against mixing old style from table1, table2 and new style from table1 inner join table2 ... in a single query. And you can alias tables using table1 as t1, shortening long table names to an easy to remember mnenomic:
select a.id
from jdc_organizations_activities a
join jdc_activity_sector as
on as.activityId = a.Id
join jdc_activity_type as at
on at.activityId = a.Id
where as.activitySector = 5
and at.activityType = 3
Or even more readable using IN:
select a.id
from jdc_organizations_activities a
where a.id in
(
select activityId
from jdc_activity_sector
where activitySector = 5
)
and a.id in
(
select activityId
from jdc_activity_type
where activityType = 3
)

How to optimize query looking for rows where conditional join rows do not exist?

I've got a table of keywords that I regularly refresh against a remote search API, and I have another table that gets a row each each time I refresh one of the keywords. I use this table to block multiple processes from stepping on each other and refreshing the same keyword, as well as stat collection. So when I spin up my program, it queries for all the keywords that don't have a request currently in process, and don't have a successful one within the last 15 mins, or whatever the interval is. All was working fine for awhile, but now the keywords_requests table has almost 2 million rows in it and things are bogging down badly. I've got indexes on almost every column in the keywords_requests table, but to no avail.
I'm logging slow queries and this one is taking forever, as you can see. What can I do?
# Query_time: 20 Lock_time: 0 Rows_sent: 568 Rows_examined: 1826718
SELECT Keyword.id, Keyword.keyword
FROM `keywords` as Keyword
LEFT JOIN `keywords_requests` as KeywordsRequest
ON (
KeywordsRequest.keyword_id = Keyword.id
AND (KeywordsRequest.status = 'success' OR KeywordsRequest.status = 'active')
AND KeywordsRequest.source_id = '29'
AND KeywordsRequest.created > FROM_UNIXTIME(1234551323)
)
WHERE KeywordsRequest.id IS NULL
GROUP BY Keyword.id
ORDER BY KeywordsRequest.created ASC;
It seems your most selective index on Keywords is one on KeywordRequest.created.
Try to rewrite query this way:
SELECT Keyword.id, Keyword.keyword
FROM `keywords` as Keyword
LEFT OUTER JOIN (
SELECT *
FROM `keywords_requests` as kr
WHERE created > FROM_UNIXTIME(1234567890) /* Happy unix_time! */
) AS KeywordsRequest
ON (
KeywordsRequest.keyword_id = Keyword.id
AND (KeywordsRequest.status = 'success' OR KeywordsRequest.status = 'active')
AND KeywordsRequest.source_id = '29'
)
WHERE keyword_id IS NULL;
It will (hopefully) hash join two not so large sources.
And Bill Karwin is right, you don't need the GROUP BY or ORDER BY
There is no fine control over the plans in MySQL, but you can try (try) to improve your query in the following ways:
Create a composite index on (keyword_id, status, source_id, created) and make it so:
SELECT Keyword.id, Keyword.keyword
FROM `keywords` as Keyword
LEFT OUTER JOIN `keywords_requests` kr
ON (
keyword_id = id
AND status = 'success'
AND source_id = '29'
AND created > FROM_UNIXTIME(1234567890)
)
WHERE keyword_id IS NULL
UNION
SELECT Keyword.id, Keyword.keyword
FROM `keywords` as Keyword
LEFT OUTER JOIN `keywords_requests` kr
ON (
keyword_id = id
AND status = 'active'
AND source_id = '29'
AND created > FROM_UNIXTIME(1234567890)
)
WHERE keyword_id IS NULL
This ideally should use NESTED LOOPS on your index.
Create a composite index on (status, source_id, created) and make it so:
SELECT Keyword.id, Keyword.keyword
FROM `keywords` as Keyword
LEFT OUTER JOIN (
SELECT *
FROM `keywords_requests` kr
WHERE
status = 'success'
AND source_id = '29'
AND created > FROM_UNIXTIME(1234567890)
UNION ALL
SELECT *
FROM `keywords_requests` kr
WHERE
status = 'active'
AND source_id = '29'
AND created > FROM_UNIXTIME(1234567890)
)
ON keyword_id = id
WHERE keyword_id IS NULL
This will hopefully use HASH JOIN on even more restricted hash table.
When diagnosing MySQL query performance, one of the first things you need to analyze is the report from EXPLAIN.
If you learn to read the information EXPLAIN gives you, then you can see where queries are failing to make use of indexes, or where they are causing expensive filesorts, or other performance red flags.
I notice in your query, the GROUP BY is irrelevant, since there will be only one NULL row returned from KeywordRequests. Also the ORDER BY is irrelevant, since you're ordering by a column that will always be NULL due to your WHERE clause. If you remove these clauses, you'll probably eliminate a filesort.
Also consider rewriting the query into other forms, and measure the performance of each. For example:
SELECT k.id, k.keyword
FROM `keywords` AS k
WHERE NOT EXISTS (
SELECT * FROM `keywords_requests` AS kr
WHERE kr.keyword_id = k.id
AND kr.status IN ('success', 'active')
AND kr.source_id = '29'
AND kr.created > FROM_UNIXTIME(1234551323)
);
Other tips:
Is kr.source_id an integer? If so, compare to the integer 29 instead of the string '29'.
Are there appropriate indexes on keyword_id, status, source_id, created? Perhaps even a compound index over all four columns would be best, since MySQL will use only one index per table in a given query.
You did a screenshot of your EXPLAIN output and posted a link in the comments. I see that the query is not using an index from Keywords, which makes sense since you're scanning every row in that table anyway. The phrase "Not exists" indicates that MySQL has optimized the LEFT OUTER JOIN a bit.
I think this should be improved over your original query. The GROUP BY/ORDER BY was probably causing it to save an intermediate data set as a temporary table, and sorting it on disk (which is very slow!). What you'd look for is "Using temporary; using filesort" in the Extra column of EXPLAIN information.
So you may have improved it enough already to mitigate the bottleneck for now.
I do notice that the possible keys probably indicate that you have individual indexes on four columns. You may be able to improve that by creating a compound index:
CREATE INDEX kr_cover ON keywords_requests
(keyword_id, created, source_id, status);
You can give MySQL a hint to use a specific index:
... FROM `keywords_requests` AS kr USE INDEX (kr_cover) WHERE ...
Dunno about MySQL but in MSSQL the lines of attack I would take are:
1) Create a covering index on KeywordsRequest status, source_id and created
2) UNION the results tog et around the OR on KeywordsRequest.status
3) Use NOT EXISTS instead o the Outer Join (and try with UNION instead of OR too)
Try this
SELECT Keyword.id, Keyword.keyword
FROM keywords as Keyword
LEFT JOIN (select * from keywords_requests where source_id = '29' and (status = 'success' OR status = 'active')
AND source_id = '29'
AND created > FROM_UNIXTIME(1234551323)
AND id IS NULL
) as KeywordsRequest
ON (
KeywordsRequest.keyword_id = Keyword.id
)
GROUP BY Keyword.id
ORDER BY KeywordsRequest.created ASC;