Using a column name in a COUNT function

Question

Using a column name in a COUNT function

Charles Kincaid

SSChampion

Points: 13593
More actions
October 14, 2013 at 9:29 pm

#281346

Comments posted to this topic are about the item Using a column name in a COUNT function
ATBCharles Kincaid

Viewing 15 posts - 1 through 15 (of 113 total)

You must be logged in to reply to this topic. Login to reply

roman.asadovsky SSC Enthusiast Points: 140 More actions · Answer 1

That is well-known behavior of the COUNT(), but the question I have is this: What makes you think that COUNT(1) in any way superior to CONT(*)?

Igor Micev SSC-Dedicated Points: 33109 More actions · Answer 2

Hi,

Why not use

select p.[rows] from sys.partitions p

where p.index_id in (0,1) and p.object_id = object_id('[schema].[TableName]')

to replace count(1)?

You already have the count for every table in sys.partitions view.

Regards,

IgorMi

Igor Micev,My blog: www.igormicev.com

danielfountain SSCarpal Tunnel Points: 4229 More actions · Answer 3

According to MS sys.partitions.rows "Indicates the approximate number of rows in this partition"

http://technet.microsoft.com/en-us/library/ms175012.aspx

Shahjahandurrani SSC Rookie Points: 42 More actions · Answer 4

Count(*) doesn't load the entire table. It uses the index to return count.

Igor Micev SSC-Dedicated Points: 33109 More actions · Answer 5

danielfountain (10/15/2013)
According to MS sys.partitions.rows "Indicates the approximate number of rows in this partition"
http://technet.microsoft.com/en-us/library/ms175012.aspx

Hi,

Microsoft recommends to use the new dynamic views instead of some deprecated for future use

The same result can be obtained by using this dynamic view as well

select i.rowcnt from sys.sysindexes i where i.id = OBJECT_ID('[schema].[TableName]')

and i.indid = 1

I often use sys.partitions and it always gives out the same result as count(1). If the maintenance of the indexes is regularly done than that info is exact. However, a good remark of you, thanks.

Regards,

IgorMi

Igor Micev,My blog: www.igormicev.com

Brandie Tarvin SSC Guru Points: 172991 More actions · Answer 6

The best way I've found for using COUNT(MyCol) is to use either the identity column or the primary key column. That way there are no NULL results to worry about.

SELECT COUNT(NameID) FROM CountTestSET ANSI_NULLS ON;

SET QUOTED_IDENTIFIER ON;

IF EXISTS (SELECT object_id

FROM sys.objects

WHERE object_id = OBJECT_ID(N'[dbo].[CountTest]')

AND type in (N'U')

)

DROP TABLE dbo.[CountTest];

CREATE TABLE dbo.[CountTest]([NameID] INT NOT NULL IDENTITY(1,1),

[Name] [nvarchar](max)

);

INSERT INTO dbo.[CountTest] ([Name]) VALUES('Sally');

INSERT INTO dbo.[CountTest] ([Name]) VALUES(NULL);

INSERT INTO dbo.[CountTest] ([Name]) VALUES('Mary');

INSERT INTO dbo.[CountTest] ([Name]) VALUES('Jane');

INSERT INTO dbo.[CountTest] ([Name]) VALUES(NULL);

INSERT INTO dbo.[CountTest] ([Name]) VALUES('Bob');

INSERT INTO dbo.[CountTest] ([Name]) VALUES('Tom');

INSERT INTO dbo.[CountTest] ([Name]) VALUES(NULL);

SELECT COUNT(NameID) FROM dbo.CountTest; --Gives count of 8

SELECT COUNT(Name) FROM dbo.CountTest; --Gives count of 5

DROP TABLE dbo.CountTest;

Brandie Tarvin, MCITP Database AdministratorLiveJournal Blog: http://brandietarvin.livejournal.com/[/url]On LinkedIn!, Google+, and Twitter.Freelance Writer: ShadowrunLatchkeys: Nevermore, Latchkeys: The Bootleg War, and Latchkeys: Roscoes in the Night are now available on Nook and Kindle.

djjunio SSC Journeyman Points: 83 More actions · Answer 7

hi, I usually have the following scenario: I have to count the distincts names

for this I use:

SELECT COUNT(distinct name) AS [COUNT distinct] FROM [CountTest];

Mighty SSCrazy Eights Points: 9063 More actions · Answer 8

...This can produce some surprising results. This is because of the way that COUNT() works...

Why the surprise if COUNT exactly does what is described in e.g. Technet?

MyDoggieJessie SSC-Forever Points: 44484 More actions · Answer 9

You know that doing COUNT(*) on a table with a lot of columns and a lot of rows can take a lot of time and memory

There is virtually no difference between SELECT(*) and SELECT(1) - the execution plans are identical, and each will produce the same number of logical reads.

______________________________________________________________________________Never argue with an idiot; Theyll drag you down to their level and beat you with experience

duncanburtenshaw Ten Centuries Points: 1323 More actions · Answer 10

MyDoggieJessie (10/15/2013)
You know that doing COUNT(*) on a table with a lot of columns and a lot of rows can take a lot of time and memory
There is virtually no difference between SELECT(*) and SELECT(1) - the execution plans are identical, and each will produce the same number of logical reads.

+1

SMeng Newbie Points: 7 More actions · Answer 11

I had learned to count on the primary key column, count(RecordID) as NumberOfRecords. However, the suggestion below to query the system tables seems to be an even better solution.

Brandie Tarvin SSC Guru Points: 172991 More actions · Answer 12

Interesting results on the execution plan. All plans but the COUNT() on Name came backup with only an expected 9B in the row size. The Name count came back much bigger (and off of what the results were).

Here are my statements:

SELECT COUNT(NameID) FROM dbo.CountTest;

SELECT COUNT(Name) FROM dbo.CountTest;

SELECT COUNT(*) FROM dbo.CountTest;

SELECT COUNT(1) FROM dbo.CountTest;

Attached is the execution plans. Identical in every way except for the selecting on the name.

Brandie Tarvin, MCITP Database AdministratorLiveJournal Blog: http://brandietarvin.livejournal.com/[/url]On LinkedIn!, Google+, and Twitter.Freelance Writer: ShadowrunLatchkeys: Nevermore, Latchkeys: The Bootleg War, and Latchkeys: Roscoes in the Night are now available on Nook and Kindle.

Jeff Gray SSChampion Points: 10667 More actions · Answer 13

I've been known to use "SELECT COUNT(*) from SomeTable with (INDEX=1)" so that I DO load the entire table.

Generally this would only be in a situation where I'm doing raw performance testing.

roman.asadovsky SSC Enthusiast Points: 140 More actions · Answer 14

MyDoggieJessie (10/15/2013)
You know that doing COUNT(*) on a table with a lot of columns and a lot of rows can take a lot of time and memory
There is virtually no difference between SELECT(*) and SELECT(1) - the execution plans are identical, and each will produce the same number of logical reads.

That's true, however SELECT * returns all the fields to the output, unlike SELECT 1, thus flooding the buffer with bytes of unnecessary data. That is not the case with COUNT(*) and COUNT(1) though. These are absolutely identical in all aspects, as far as I know.

Please correct me if I am wrong - I would love to learn if there is a difference.