问题
What is the main purpose of using CROSS APPLY?
I have read (vaguely, through posts on the Internet) that cross apply can be more efficient when selecting over large data sets if you are partitioning. (Paging comes to mind)
I also know that CROSS APPLY doesn\'t require a UDF as the right-table.
In most INNER JOIN queries (one-to-many relationships), I could rewrite them to use CROSS APPLY, but they always give me equivalent execution plans.
Can anyone give me a good example of when CROSS APPLY makes a difference in those cases where INNER JOIN will work as well?
Edit:
Here\'s a trivial example, where the execution plans are exactly the same. (Show me one where they differ and where cross apply is faster/more efficient)
create table Company (
companyId int identity(1,1)
, companyName varchar(100)
, zipcode varchar(10)
, constraint PK_Company primary key (companyId)
)
GO
create table Person (
personId int identity(1,1)
, personName varchar(100)
, companyId int
, constraint FK_Person_CompanyId foreign key (companyId) references dbo.Company(companyId)
, constraint PK_Person primary key (personId)
)
GO
insert Company
select \'ABC Company\', \'19808\' union
select \'XYZ Company\', \'08534\' union
select \'123 Company\', \'10016\'
insert Person
select \'Alan\', 1 union
select \'Bobby\', 1 union
select \'Chris\', 1 union
select \'Xavier\', 2 union
select \'Yoshi\', 2 union
select \'Zambrano\', 2 union
select \'Player 1\', 3 union
select \'Player 2\', 3 union
select \'Player 3\', 3
/* using CROSS APPLY */
select *
from Person p
cross apply (
select *
from Company c
where p.companyid = c.companyId
) Czip
/* the equivalent query using INNER JOIN */
select *
from Person p
inner join Company c on p.companyid = c.companyId
回答1:
Can anyone give me a good example of when CROSS APPLY makes a difference in those cases where INNER JOIN will work as well?
See the article in my blog for detailed performance comparison:
- INNER JOIN vs. CROSS APPLY
CROSS APPLY works better on things that have no simple JOIN condition.
This one selects 3 last records from t2 for each record from t1:
SELECT t1.*, t2o.*
FROM t1
CROSS APPLY
(
SELECT TOP 3 *
FROM t2
WHERE t2.t1_id = t1.id
ORDER BY
t2.rank DESC
) t2o
It cannot be easily formulated with an INNER JOIN condition.
You could probably do something like that using CTE's and window function:
WITH t2o AS
(
SELECT t2.*, ROW_NUMBER() OVER (PARTITION BY t1_id ORDER BY rank) AS rn
FROM t2
)
SELECT t1.*, t2o.*
FROM t1
INNER JOIN
t2o
ON t2o.t1_id = t1.id
AND t2o.rn <= 3
, but this is less readable and probably less efficient.
Update:
Just checked.
master is a table of about 20,000,000 records with a PRIMARY KEY on id.
This query:
WITH q AS
(
SELECT *, ROW_NUMBER() OVER (ORDER BY id) AS rn
FROM master
),
t AS
(
SELECT 1 AS id
UNION ALL
SELECT 2
)
SELECT *
FROM t
JOIN q
ON q.rn <= t.id
runs for almost 30 seconds, while this one:
WITH t AS
(
SELECT 1 AS id
UNION ALL
SELECT 2
)
SELECT *
FROM t
CROSS APPLY
(
SELECT TOP (t.id) m.*
FROM master m
ORDER BY
id
) q
is instant.
回答2:
cross apply sometimes enables you to do things that you cannot do with inner join.
Example (a syntax error):
select F.* from sys.objects O
inner join dbo.myTableFun(O.name) F
on F.schema_id= O.schema_id
This is a syntax error, because, when used with inner join, table functions can only take variables or constants as parameters. (I.e., the table function parameter cannot depend on another table's column.)
However:
select F.* from sys.objects O
cross apply ( select * from dbo.myTableFun(O.name) ) F
where F.schema_id= O.schema_id
This is legal.
Edit: Or alternatively, shorter syntax: (by ErikE)
select F.* from sys.objects O
cross apply dbo.myTableFun(O.name) F
where F.schema_id= O.schema_id
Edit:
Note: Informix 12.10 xC2+ has Lateral Derived Tables and Postgresql (9.3+) has Lateral Subqueries which can be used to a similar effect.
回答3:
Consider you have two tables.
MASTER TABLE
x------x--------------------x
| Id | Name |
x------x--------------------x
| 1 | A |
| 2 | B |
| 3 | C |
x------x--------------------x
DETAILS TABLE
x------x--------------------x-------x
| Id | PERIOD | QTY |
x------x--------------------x-------x
| 1 | 2014-01-13 | 10 |
| 1 | 2014-01-11 | 15 |
| 1 | 2014-01-12 | 20 |
| 2 | 2014-01-06 | 30 |
| 2 | 2014-01-08 | 40 |
x------x--------------------x-------x
There are many situations where we need to replace INNER JOIN with CROSS APPLY.
1. Join two tables based on TOP n results
Consider if we need to select Id and Name from Master and last two dates for each Id from Details table.
SELECT M.ID,M.NAME,D.PERIOD,D.QTY
FROM MASTER M
INNER JOIN
(
SELECT TOP 2 ID, PERIOD,QTY
FROM DETAILS D
ORDER BY CAST(PERIOD AS DATE)DESC
)D
ON M.ID=D.ID
- SQL FIDDLE
The above query generates the following result.
x------x---------x--------------x-------x
| Id | Name | PERIOD | QTY |
x------x---------x--------------x-------x
| 1 | A | 2014-01-13 | 10 |
| 1 | A | 2014-01-12 | 20 |
x------x---------x--------------x-------x
See, it generated results for last two dates with last two date's Id and then joined these records only in the outer query on Id, which is wrong. To accomplish this, we need to use CROSS APPLY.
SELECT M.ID,M.NAME,D.PERIOD,D.QTY
FROM MASTER M
CROSS APPLY
(
SELECT TOP 2 ID, PERIOD,QTY
FROM DETAILS D
WHERE M.ID=D.ID
ORDER BY CAST(PERIOD AS DATE)DESC
)D
- SQL FIDDLE
and forms the following result.
x------x---------x--------------x-------x
| Id | Name | PERIOD | QTY |
x------x---------x--------------x-------x
| 1 | A | 2014-01-13 | 10 |
| 1 | A | 2014-01-12 | 20 |
| 2 | B | 2014-01-08 | 40 |
| 2 | B | 2014-01-06 | 30 |
x------x---------x--------------x-------x
Here's how it works. The query inside CROSS APPLY can reference the outer table, where INNER JOIN cannot do this (it throws compile error). When finding the last two dates, joining is done inside CROSS APPLY i.e., WHERE M.ID=D.ID.
2. When we need INNER JOIN functionality using functions.
CROSS APPLY can be used as a replacement with INNER JOIN when we need to get result from Master table and a function.
SELECT M.ID,M.NAME,C.PERIOD,C.QTY
FROM MASTER M
CROSS APPLY dbo.FnGetQty(M.ID) C
And here is the function
CREATE FUNCTION FnGetQty
(
@Id INT
)
RETURNS TABLE
AS
RETURN
(
SELECT ID,PERIOD,QTY
FROM DETAILS
WHERE ID=@Id
)
- SQL FIDDLE
which generated the following result
x------x---------x--------------x-------x
| Id | Name | PERIOD | QTY |
x------x---------x--------------x-------x
| 1 | A | 2014-01-13 | 10 |
| 1 | A | 2014-01-11 | 15 |
| 1 | A | 2014-01-12 | 20 |
| 2 | B | 2014-01-06 | 30 |
| 2 | B | 2014-01-08 | 40 |
x------x---------x--------------x-------x
ADDITIONAL ADVANTAGE OF CROSS APPLY
APPLY can be used as a replacement for UNPIVOT. Either CROSS APPLY or OUTER APPLY can be used here, which are interchangeable.
Consider you have the below table(named MYTABLE).
x------x-------------x--------------x
| Id | FROMDATE | TODATE |
x------x-------------x--------------x
| 1 | 2014-01-11 | 2014-01-13 |
| 1 | 2014-02-23 | 2014-02-27 |
| 2 | 2014-05-06 | 2014-05-30 |
| 3 | NULL | NULL |
x------x-------------x--------------x
The query is below.
SELECT DISTINCT ID,DATES
FROM MYTABLE
CROSS APPLY(VALUES (FROMDATE),(TODATE))
COLUMNNAMES(DATES)
- SQL FIDDLE
which brings you the result
x------x-------------x
| Id | DATES |
x------x-------------x
| 1 | 2014-01-11 |
| 1 | 2014-01-13 |
| 1 | 2014-02-23 |
| 1 | 2014-02-27 |
| 2 | 2014-05-06 |
| 2 | 2014-05-30 |
| 3 | NULL |
x------x-------------x
回答4:
It seems to me that CROSS APPLY can fill a certain gap when working with calculated fields in complex/nested queries, and make them simpler and more readable.
Simple example: you have a DoB and you want to present multiple age-related fields that will also rely on other data sources (such as employment), like Age, AgeGroup, AgeAtHiring, MinimumRetirementDate, etc. for use in your end-user application (Excel PivotTables, for example).
Options are limited and rarely elegant:
JOIN subqueries cannot introduce new values in the dataset based on data in the parent query (it must stand on its own).
UDFs are neat, but slow as they tend to prevent parallel operations. And being a separate entity can be a good (less code) or a bad (where is the code) thing.
Junction tables. Sometimes they can work, but soon enough you're joining subqueries with tons of UNIONs. Big mess.
Create yet another single-purpose view, assuming your calculations don't require data obtained mid-way through your main query.
Intermediary tables. Yes... that usually works, and often a good option as they can be indexed and fast, but performance can also drop due to to UPDATE statements not being parallel and not allowing to cascade formulas (reuse results) to update several fields within the same statement. And sometimes you'd just prefer to do things in one pass.
Nesting queries. Yes at any point you can put parenthesis on your entire query and use it as a subquery upon which you can manipulate source data and calculated fields alike. But you can only do this so much before it gets ugly. Very ugly.
Repeating code. What is the greatest value of 3 long (CASE...ELSE...END) statements? That's gonna be readable!
- Tell your clients to calculate the damn things themselves.
Did I miss something? Probably, so feel free to comment. But hey, CROSS APPLY is like a godsend in such situations: you just add a simple CROSS APPLY (select tbl.value + 1 as someFormula) as crossTbl and voilà! Your new field is now ready for use practically like it had always been there in your source data.
Values introduced through CROSS APPLY can...
- be used to create one or multiple calculated fields without adding performance, complexity or readability issues to the mix
- like with JOINs, several subsequent CROSS APPLY statements can refer to themselves:
CROSS APPLY (select crossTbl.someFormula + 1 as someMoreFormula) as crossTbl2 - you can use values introduced by a CROSS APPLY in subsequent JOIN conditions
- As a bonus, there's the Table-valued function aspect
Dang, there's nothing they can't do!
回答5:
Cross apply works well with an XML field as well. If you wish to select node values in combination with other fields.
For example, if you have a table containing some xml
<root> <subnode1> <some_node value="1" /> <some_node value="2" /> <some_node value="3" /> <some_node value="4" /> </subnode1> </root>
Using the query
SELECT
id as [xt_id]
,xmlfield.value('(/root/@attribute)[1]', 'varchar(50)') root_attribute_value
,node_attribute_value = [some_node].value('@value', 'int')
,lt.lt_name
FROM dbo.table_with_xml xt
CROSS APPLY xmlfield.nodes('/root/subnode1/some_node') as g ([some_node])
LEFT OUTER JOIN dbo.lookup_table lt
ON [some_node].value('@value', 'int') = lt.lt_id
Will return a result
xt_id root_attribute_value node_attribute_value lt_name
----------------------------------------------------------------------
1 test1 1 Benefits
1 test1 4 FINRPTCOMPANY
回答6:
This has already been answered very well technically, but let me give a concrete example of how it's extremely useful:
Lets say you have two tables, Customer and Order. Customers have many Orders.
I want to create a view that gives me details about customers, and the most recent order they've made. With just JOINS, this would require some self-joins and aggregation which isn't pretty. But with Cross Apply, its super easy:
SELECT *
FROM Customer
CROSS APPLY (
SELECT TOP 1 *
FROM Order
WHERE Order.CustomerId = Customer.CustomerId
ORDER BY OrderDate DESC
) T
回答7:
Cross apply can be used to replace subquery's where you need a column of the subquery
subquery
select * from person p where
p.companyId in(select c.companyId from company c where c.companyname like '%yyy%')
here i won't be able to select the columns of company table so, using cross apply
select P.*,T.CompanyName
from Person p
cross apply (
select *
from Company C
where p.companyid = c.companyId and c.CompanyName like '%yyy%'
) T
回答8:
I guess it should be readability ;)
CROSS APPLY will be somewhat unique for people reading to tell them that a UDF is being used which will be applied to each row from the table on the left.
Ofcourse, there are other limitations where a CROSS APPLY is better used than JOIN which other friends have posted above.
回答9:
Here is an article that explains it all, with their performance difference and usage over JOINS.
SQL Server CROSS APPLY and OUTER APPLY over JOINS
As suggested in this article, there is no performance difference between them for normal join operations (INNER AND CROSS).
The usage difference arrives when you have to do a query like this:
CREATE FUNCTION dbo.fn_GetAllEmployeeOfADepartment(@DeptID AS INT)
RETURNS TABLE
AS
RETURN
(
SELECT * FROM Employee E
WHERE E.DepartmentID = @DeptID
)
GO
SELECT * FROM Department D
CROSS APPLY dbo.fn_GetAllEmployeeOfADepartment(D.DepartmentID)
That is, when you have to relate with function. This cannot be done using INNER JOIN, which would give you the error "The multi-part identifier "D.DepartmentID" could not be bound." Here the value is passed to the function as each row is read. Sounds cool to me. :)
回答10:
Well I am not sure if this qualifies as a reason to use Cross Apply versus Inner Join, but this query was answered for me in a Forum Post using Cross Apply, so I am not sure if there is an equalivent method using Inner Join:
Create PROCEDURE [dbo].[Message_FindHighestMatches]
-- Declare the Topical Neighborhood
@TopicalNeighborhood nchar(255)
AS BEGIN
-- SET NOCOUNT ON added to prevent extra result sets from
-- interfering with SELECT statements.
SET NOCOUNT ON
Create table #temp
(
MessageID int,
Subjects nchar(255),
SubjectsCount int
)
Insert into #temp Select MessageID, Subjects, SubjectsCount From Message
Select Top 20 MessageID, Subjects, SubjectsCount,
(t.cnt * 100)/t3.inputvalues as MatchPercentage
From #temp
cross apply (select count(*) as cnt from dbo.Split(Subjects,',') as t1
join dbo.Split(@TopicalNeighborhood,',') as t2
on t1.value = t2.value) as t
cross apply (select count(*) as inputValues from dbo.Split(@TopicalNeighborhood,',')) as t3
Order By MatchPercentage desc
drop table #temp
END
回答11:
The essence of the APPLY operator is to allow correlation between left and right side of the operator in the FROM clause.
In contrast to JOIN, the correlation between inputs is not allowed.
Speaking about correlation in APPLY operator, I mean on the right hand side we can put:
- a derived table - as a correlated subquery with an alias
- a table valued function - a conceptual view with parameters, where the parameter can refer to the left side
Both can return multiple columns and rows.
回答12:
This is perhaps an old question, but I still love the power of CROSS APPLY to simplify the re-use of logic and to provide a "chaining" mechanism for results.
I've provided a SQL Fiddle below which shows a simple example of how you can use CROSS APPLY to perform complex logical operations on your data set without things getting at all messy. It's not hard to extrapolate from here more complex calculations.
http://sqlfiddle.com/#!3/23862/2
来源：https://stackoverflow.com/questions/1139160/when-should-i-use-cross-apply-over-inner-join