Related
Is there a difference between
1st Way
Alter Table Test_Table
Add Test_Column
and
2nd way
Alter Table Test_Table
Add Column Test_Column
Alter Table Test_Table Add Test_Column
Use for Add new columns in table
Alter Table Test_Table Add Column Test_Column
use for Modify column in table
It depends on the database.
For example Oracle enough
ALTER TABLE xxx ADD column_name TYPE;
But in postgresql you must column after add
ALTER TABLE xxx ADD COLUMN column_name TYPE;
We have a 5GB table (nearly 500 million rows) and we want to remove the identity property on one of the column, but when we try to do this through SSMS - it times out.
Can this be done through T-SQL?
You cannot remove an IDENTITY specification once set.
To remove the entire column:
ALTER TABLE yourTable
DROP COLUMN yourCOlumn;
Information about ALTER TABLE here
If you need to keep the data, but remove the IDENTITY column, you will need to:
Create a new column
Transfer the data from the existing IDENTITY column to the new column
Drop the existing IDENTITY column.
Rename the new column to the original column name
If you want to do this without adding and populating a new column, without reordering the columns, and with almost no downtime because no data is changing on the table, let's do some magic with partitioning functionality (but since no partitions are used you don't need Enterprise edition):
Remove all foreign keys that point to this table
Script the table to be created; rename everything e.g. 'MyTable2', 'MyIndex2', etc.
Remove the IDENTITY specification.
You should now have two "identical"-ish tables, one full, the other empty with no IDENTITY.
Run ALTER TABLE [Original] SWITCH TO [Original2]
Now your original table will be empty and the new one will have the data. You have switched the metadata for the two tables (instant).
Drop the original (now-empty table), exec sys.sp_rename to rename the various schema objects back to the original names, and then you can recreate your foreign keys.
For example, given:
CREATE TABLE Original
(
Id INT IDENTITY PRIMARY KEY
, Value NVARCHAR(300)
);
CREATE NONCLUSTERED INDEX IX_Original_Value ON Original (Value);
INSERT INTO Original
SELECT 'abcd'
UNION ALL
SELECT 'defg';
You can do the following:
--create new table with no IDENTITY
CREATE TABLE Original2
(
Id INT PRIMARY KEY
, Value NVARCHAR(300)
);
CREATE NONCLUSTERED INDEX IX_Original_Value2 ON Original2 (Value);
--data before switch
SELECT 'Original', *
FROM Original
UNION ALL
SELECT 'Original2', *
FROM Original2;
ALTER TABLE Original SWITCH TO Original2;
--data after switch
SELECT 'Original', *
FROM Original
UNION ALL
SELECT 'Original2', *
FROM Original2;
--clean up
IF NOT EXISTS (SELECT * FROM Original) DROP TABLE Original;
EXEC sys.sp_rename 'Original2.IX_Original_Value2', 'IX_Original_Value', 'INDEX';
EXEC sys.sp_rename 'Original2', 'Original', 'OBJECT';
UPDATE Original
SET Id = Id + 1;
SELECT *
FROM Original;
For the record, as this has become increasingly popular, I have wanted to track down the original source (I didn't think of it myself), but of course I've long forgotten where I found the original idea. It may have been here; this was the only one I could find predating this answer (time flies, boys and girls): https://social.technet.microsoft.com/wiki/contents/articles/17738.sql-server-quick-way-to-remove-the-identity-property.aspx
This gets messy with foreign and primary key constraints, so here's some scripts to help you on your way:
First, create a duplicate column with a temporary name:
alter table yourTable add tempId int NOT NULL default -1;
update yourTable set tempId = id;
Next, get the name of your primary key constraint:
SELECT * FROM INFORMATION_SCHEMA.TABLE_CONSTRAINTS WHERE TABLE_NAME = 'yourTable';
Now try drop the primary key constraint for your column:
ALTER TABLE yourTable DROP CONSTRAINT PK_yourTable_id;
If you have foreign keys, it will fail, so if so drop the foreign key constraints. KEEP TRACK OF WHICH TABLES YOU RUN THIS FOR SO YOU CAN ADD THE CONSTRAINTS BACK IN LATER!!!
SELECT * FROM INFORMATION_SCHEMA.TABLE_CONSTRAINTS WHERE TABLE_NAME = 'otherTable';
alter table otherTable drop constraint fk_otherTable_yourTable;
commit;
..
Once all of your foreign key constraints have been removed, you'll be able to remove the PK constraint, drop that column, rename your temp column, and add the PK constraint to that column:
ALTER TABLE yourTable DROP CONSTRAINT PK_yourTable_id;
alter table yourTable drop column id;
EXEC sp_rename 'yourTable.tempId', 'id', 'COLUMN';
ALTER TABLE yourTable ADD CONSTRAINT PK_yourTable_id PRIMARY KEY (id)
commit;
Finally, add the FK constraints back in:
alter table otherTable add constraint fk_otherTable_yourTable foreign key (yourTable_id) references yourTable(id);
..
El Fin!
I just had this same problem. 4 statements in SSMS instead of using the GUI and it was very fast.
Make a new column
alter table users add newusernum int;
Copy values over
update users set newusernum=usernum;
Drop the old column
alter table users drop column usernum;
Rename the new column to the old column name
EXEC sp_RENAME 'users.newusernum' , 'usernum', 'COLUMN';
Following script removes Identity field for a column named 'Id'
Hope it helps.
BEGIN TRAN
BEGIN TRY
EXEC sp_rename '[SomeTable].[Id]', 'OldId';
ALTER TABLE [SomeTable] ADD Id int NULL
EXEC ('UPDATE [SomeTable] SET Id = OldId')
ALTER TABLE [SomeTable] NOCHECK CONSTRAINT ALL
ALTER TABLE [SomeTable] DROP CONSTRAINT [PK_constraintName];
ALTER TABLE [SomeTable] DROP COLUMN OldId
ALTER TABLE [SomeTable] ALTER COLUMN [Id] INTEGER NOT NULL
ALTER TABLE [SomeTable] ADD CONSTRAINT PK_JobInfo PRIMARY KEY (Id)
ALTER TABLE [SomeTable] CHECK CONSTRAINT ALL
COMMIT TRAN
END TRY
BEGIN CATCH
ROLLBACK TRAN
SELECT ERROR_MESSAGE ()
END CATCH
Bellow code working as fine, when we don't know identity column name.
Need to copy data into new temp table like Invoice_DELETED.
and next time we using:
insert into Invoice_DELETED select * from Invoice where ...
SELECT t1.*
INTO Invoice_DELETED
FROM Invoice t1
LEFT JOIN Invoice ON 1 = 0
--WHERE t1.InvoiceID = #InvoiceID
For more explanation see:
https://dba.stackexchange.com/a/138345/101038
ALTER TABLE tablename add newcolumn int
update tablename set newcolumn=existingcolumnname
ALTER TABLE tablename DROP COLUMN existingcolumnname;
EXEC sp_RENAME 'tablename.oldcolumn' , 'newcolumnname', 'COLUMN'
However above code works only if no primary-foreign key relation
In SQL Server you can turn on and off identity insert like this:
SET IDENTITY_INSERT table_name ON
-- run your queries here
SET IDENTITY_INSERT table_name OFF
Just for someone who have the same problem I did.
If you just want to make some insert just once you can do something like this.
Lets suppose you have a table with two columns
ID Identity (1,1) | Name Varchar
and want to insert a row with the ID = 4. So you Reseed it to 3 so the next one is 4
DBCC CHECKIDENT([YourTable], RESEED, 3)
Make the Insert
INSERT INTO [YourTable]
( Name )
VALUES ( 'Client' )
And get your seed back to the highest ID, lets suppose is 15
DBCC CHECKIDENT([YourTable], RESEED, 15)
Done!
I had the same requirement, and you could try this way, which I personally recommend you, please manually design your table and generate the script, and what I did below was renaming the old table and also its constraint for backup.
/* To prevent any potential data loss issues, you should review this script in detail before running it outside the context of the database designer.*/
BEGIN TRANSACTION
SET QUOTED_IDENTIFIER ON
SET ARITHABORT ON
SET NUMERIC_ROUNDABORT OFF
SET CONCAT_NULL_YIELDS_NULL ON
SET ANSI_NULLS ON
SET ANSI_PADDING ON
SET ANSI_WARNINGS ON
COMMIT
BEGIN TRANSACTION
GO
ALTER TABLE dbo.SI_Provider_Profile
DROP CONSTRAINT DF_SI_Provider_Profile_SIdtDateTimeStamp
GO
ALTER TABLE dbo.SI_Provider_Profile
DROP CONSTRAINT DF_SI_Provider_Profile_SIbHotelPreLoaded
GO
CREATE TABLE dbo.Tmp_SI_Provider_Profile
(
SI_lProvider_Profile_ID int NOT NULL,
SI_lSerko_Integrator_Token_ID int NOT NULL,
SI_sSerko_Integrator_Provider varchar(50) NOT NULL,
SI_sSerko_Integrator_Profile varchar(50) NOT NULL,
SI_dtDate_Time_Stamp datetime NOT NULL,
SI_lProvider_ID int NULL,
SI_sDisplay_Name varchar(10) NULL,
SI_lPurchased_From int NULL,
SI_sProvider_UniqueID varchar(255) NULL,
SI_bHotel_Pre_Loaded bit NOT NULL,
SI_sSiteName varchar(255) NULL
) ON [PRIMARY]
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile SET (LOCK_ESCALATION = TABLE)
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile ADD CONSTRAINT
DF_SI_Provider_Profile_SIdtDateTimeStamp DEFAULT (getdate()) FOR SI_dtDate_Time_Stamp
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile ADD CONSTRAINT
DF_SI_Provider_Profile_SIbHotelPreLoaded DEFAULT ((0)) FOR SI_bHotel_Pre_Loaded
GO
IF EXISTS(SELECT * FROM dbo.SI_Provider_Profile)
EXEC('INSERT INTO dbo.Tmp_SI_Provider_Profile (SI_lProvider_Profile_ID, SI_lSerko_Integrator_Token_ID, SI_sSerko_Integrator_Provider, SI_sSerko_Integrator_Profile, SI_dtDate_Time_Stamp, SI_lProvider_ID, SI_sDisplay_Name, SI_lPurchased_From, SI_sProvider_UniqueID, SI_bHotel_Pre_Loaded, SI_sSiteName)
SELECT SI_lProvider_Profile_ID, SI_lSerko_Integrator_Token_ID, SI_sSerko_Integrator_Provider, SI_sSerko_Integrator_Profile, SI_dtDate_Time_Stamp, SI_lProvider_ID, SI_sDisplay_Name, SI_lPurchased_From, SI_sProvider_UniqueID, SI_bHotel_Pre_Loaded, SI_sSiteName FROM dbo.SI_Provider_Profile WITH (HOLDLOCK TABLOCKX)')
GO
-- Rename the primary key constraint or unique key In SQL Server constraints such as primary keys or foreign keys are objects in their own right, even though they are dependent upon the "containing" table.
EXEC sp_rename 'dbo.SI_Provider_Profile.PK_SI_Provider_Profile', 'PK_SI_Provider_Profile_Old';
GO
-- backup old table in case of
EXECUTE sp_rename N'dbo.SI_Provider_Profile', N'SI_Provider_Profile_Old', 'OBJECT'
GO
EXECUTE sp_rename N'dbo.Tmp_SI_Provider_Profile', N'SI_Provider_Profile', 'OBJECT'
GO
ALTER TABLE dbo.SI_Provider_Profile ADD CONSTRAINT
PK_SI_Provider_Profile PRIMARY KEY NONCLUSTERED
(
SI_lProvider_Profile_ID
) WITH( PAD_INDEX = OFF, FILLFACTOR = 90, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
GO
COMMIT TRANSACTION
ALTER TABLE TABLE_NAME MODIFY (COLUMN_NAME DROP IDENTITY);
How do I remove AUTO_INCREMENT of a column in DB2?
I tried
ALTER TABLE my_table ALTER my_id
but no luck. What is the correct SQL statement?
You can modify a column definition of a DB2 table and remove the generation of values:
ALTER TABLE mytable ALTER COLUMN mycol DROP GENERATED;
You can remove the Auto_increment by using the following
ALTER TABLE table_name ALTER COLUMN column_name DROP IDENTITY;
We have a 5GB table (nearly 500 million rows) and we want to remove the identity property on one of the column, but when we try to do this through SSMS - it times out.
Can this be done through T-SQL?
You cannot remove an IDENTITY specification once set.
To remove the entire column:
ALTER TABLE yourTable
DROP COLUMN yourCOlumn;
Information about ALTER TABLE here
If you need to keep the data, but remove the IDENTITY column, you will need to:
Create a new column
Transfer the data from the existing IDENTITY column to the new column
Drop the existing IDENTITY column.
Rename the new column to the original column name
If you want to do this without adding and populating a new column, without reordering the columns, and with almost no downtime because no data is changing on the table, let's do some magic with partitioning functionality (but since no partitions are used you don't need Enterprise edition):
Remove all foreign keys that point to this table
Script the table to be created; rename everything e.g. 'MyTable2', 'MyIndex2', etc.
Remove the IDENTITY specification.
You should now have two "identical"-ish tables, one full, the other empty with no IDENTITY.
Run ALTER TABLE [Original] SWITCH TO [Original2]
Now your original table will be empty and the new one will have the data. You have switched the metadata for the two tables (instant).
Drop the original (now-empty table), exec sys.sp_rename to rename the various schema objects back to the original names, and then you can recreate your foreign keys.
For example, given:
CREATE TABLE Original
(
Id INT IDENTITY PRIMARY KEY
, Value NVARCHAR(300)
);
CREATE NONCLUSTERED INDEX IX_Original_Value ON Original (Value);
INSERT INTO Original
SELECT 'abcd'
UNION ALL
SELECT 'defg';
You can do the following:
--create new table with no IDENTITY
CREATE TABLE Original2
(
Id INT PRIMARY KEY
, Value NVARCHAR(300)
);
CREATE NONCLUSTERED INDEX IX_Original_Value2 ON Original2 (Value);
--data before switch
SELECT 'Original', *
FROM Original
UNION ALL
SELECT 'Original2', *
FROM Original2;
ALTER TABLE Original SWITCH TO Original2;
--data after switch
SELECT 'Original', *
FROM Original
UNION ALL
SELECT 'Original2', *
FROM Original2;
--clean up
IF NOT EXISTS (SELECT * FROM Original) DROP TABLE Original;
EXEC sys.sp_rename 'Original2.IX_Original_Value2', 'IX_Original_Value', 'INDEX';
EXEC sys.sp_rename 'Original2', 'Original', 'OBJECT';
UPDATE Original
SET Id = Id + 1;
SELECT *
FROM Original;
For the record, as this has become increasingly popular, I have wanted to track down the original source (I didn't think of it myself), but of course I've long forgotten where I found the original idea. It may have been here; this was the only one I could find predating this answer (time flies, boys and girls): https://social.technet.microsoft.com/wiki/contents/articles/17738.sql-server-quick-way-to-remove-the-identity-property.aspx
This gets messy with foreign and primary key constraints, so here's some scripts to help you on your way:
First, create a duplicate column with a temporary name:
alter table yourTable add tempId int NOT NULL default -1;
update yourTable set tempId = id;
Next, get the name of your primary key constraint:
SELECT * FROM INFORMATION_SCHEMA.TABLE_CONSTRAINTS WHERE TABLE_NAME = 'yourTable';
Now try drop the primary key constraint for your column:
ALTER TABLE yourTable DROP CONSTRAINT PK_yourTable_id;
If you have foreign keys, it will fail, so if so drop the foreign key constraints. KEEP TRACK OF WHICH TABLES YOU RUN THIS FOR SO YOU CAN ADD THE CONSTRAINTS BACK IN LATER!!!
SELECT * FROM INFORMATION_SCHEMA.TABLE_CONSTRAINTS WHERE TABLE_NAME = 'otherTable';
alter table otherTable drop constraint fk_otherTable_yourTable;
commit;
..
Once all of your foreign key constraints have been removed, you'll be able to remove the PK constraint, drop that column, rename your temp column, and add the PK constraint to that column:
ALTER TABLE yourTable DROP CONSTRAINT PK_yourTable_id;
alter table yourTable drop column id;
EXEC sp_rename 'yourTable.tempId', 'id', 'COLUMN';
ALTER TABLE yourTable ADD CONSTRAINT PK_yourTable_id PRIMARY KEY (id)
commit;
Finally, add the FK constraints back in:
alter table otherTable add constraint fk_otherTable_yourTable foreign key (yourTable_id) references yourTable(id);
..
El Fin!
I just had this same problem. 4 statements in SSMS instead of using the GUI and it was very fast.
Make a new column
alter table users add newusernum int;
Copy values over
update users set newusernum=usernum;
Drop the old column
alter table users drop column usernum;
Rename the new column to the old column name
EXEC sp_RENAME 'users.newusernum' , 'usernum', 'COLUMN';
Following script removes Identity field for a column named 'Id'
Hope it helps.
BEGIN TRAN
BEGIN TRY
EXEC sp_rename '[SomeTable].[Id]', 'OldId';
ALTER TABLE [SomeTable] ADD Id int NULL
EXEC ('UPDATE [SomeTable] SET Id = OldId')
ALTER TABLE [SomeTable] NOCHECK CONSTRAINT ALL
ALTER TABLE [SomeTable] DROP CONSTRAINT [PK_constraintName];
ALTER TABLE [SomeTable] DROP COLUMN OldId
ALTER TABLE [SomeTable] ALTER COLUMN [Id] INTEGER NOT NULL
ALTER TABLE [SomeTable] ADD CONSTRAINT PK_JobInfo PRIMARY KEY (Id)
ALTER TABLE [SomeTable] CHECK CONSTRAINT ALL
COMMIT TRAN
END TRY
BEGIN CATCH
ROLLBACK TRAN
SELECT ERROR_MESSAGE ()
END CATCH
Bellow code working as fine, when we don't know identity column name.
Need to copy data into new temp table like Invoice_DELETED.
and next time we using:
insert into Invoice_DELETED select * from Invoice where ...
SELECT t1.*
INTO Invoice_DELETED
FROM Invoice t1
LEFT JOIN Invoice ON 1 = 0
--WHERE t1.InvoiceID = #InvoiceID
For more explanation see:
https://dba.stackexchange.com/a/138345/101038
ALTER TABLE tablename add newcolumn int
update tablename set newcolumn=existingcolumnname
ALTER TABLE tablename DROP COLUMN existingcolumnname;
EXEC sp_RENAME 'tablename.oldcolumn' , 'newcolumnname', 'COLUMN'
However above code works only if no primary-foreign key relation
In SQL Server you can turn on and off identity insert like this:
SET IDENTITY_INSERT table_name ON
-- run your queries here
SET IDENTITY_INSERT table_name OFF
Just for someone who have the same problem I did.
If you just want to make some insert just once you can do something like this.
Lets suppose you have a table with two columns
ID Identity (1,1) | Name Varchar
and want to insert a row with the ID = 4. So you Reseed it to 3 so the next one is 4
DBCC CHECKIDENT([YourTable], RESEED, 3)
Make the Insert
INSERT INTO [YourTable]
( Name )
VALUES ( 'Client' )
And get your seed back to the highest ID, lets suppose is 15
DBCC CHECKIDENT([YourTable], RESEED, 15)
Done!
I had the same requirement, and you could try this way, which I personally recommend you, please manually design your table and generate the script, and what I did below was renaming the old table and also its constraint for backup.
/* To prevent any potential data loss issues, you should review this script in detail before running it outside the context of the database designer.*/
BEGIN TRANSACTION
SET QUOTED_IDENTIFIER ON
SET ARITHABORT ON
SET NUMERIC_ROUNDABORT OFF
SET CONCAT_NULL_YIELDS_NULL ON
SET ANSI_NULLS ON
SET ANSI_PADDING ON
SET ANSI_WARNINGS ON
COMMIT
BEGIN TRANSACTION
GO
ALTER TABLE dbo.SI_Provider_Profile
DROP CONSTRAINT DF_SI_Provider_Profile_SIdtDateTimeStamp
GO
ALTER TABLE dbo.SI_Provider_Profile
DROP CONSTRAINT DF_SI_Provider_Profile_SIbHotelPreLoaded
GO
CREATE TABLE dbo.Tmp_SI_Provider_Profile
(
SI_lProvider_Profile_ID int NOT NULL,
SI_lSerko_Integrator_Token_ID int NOT NULL,
SI_sSerko_Integrator_Provider varchar(50) NOT NULL,
SI_sSerko_Integrator_Profile varchar(50) NOT NULL,
SI_dtDate_Time_Stamp datetime NOT NULL,
SI_lProvider_ID int NULL,
SI_sDisplay_Name varchar(10) NULL,
SI_lPurchased_From int NULL,
SI_sProvider_UniqueID varchar(255) NULL,
SI_bHotel_Pre_Loaded bit NOT NULL,
SI_sSiteName varchar(255) NULL
) ON [PRIMARY]
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile SET (LOCK_ESCALATION = TABLE)
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile ADD CONSTRAINT
DF_SI_Provider_Profile_SIdtDateTimeStamp DEFAULT (getdate()) FOR SI_dtDate_Time_Stamp
GO
ALTER TABLE dbo.Tmp_SI_Provider_Profile ADD CONSTRAINT
DF_SI_Provider_Profile_SIbHotelPreLoaded DEFAULT ((0)) FOR SI_bHotel_Pre_Loaded
GO
IF EXISTS(SELECT * FROM dbo.SI_Provider_Profile)
EXEC('INSERT INTO dbo.Tmp_SI_Provider_Profile (SI_lProvider_Profile_ID, SI_lSerko_Integrator_Token_ID, SI_sSerko_Integrator_Provider, SI_sSerko_Integrator_Profile, SI_dtDate_Time_Stamp, SI_lProvider_ID, SI_sDisplay_Name, SI_lPurchased_From, SI_sProvider_UniqueID, SI_bHotel_Pre_Loaded, SI_sSiteName)
SELECT SI_lProvider_Profile_ID, SI_lSerko_Integrator_Token_ID, SI_sSerko_Integrator_Provider, SI_sSerko_Integrator_Profile, SI_dtDate_Time_Stamp, SI_lProvider_ID, SI_sDisplay_Name, SI_lPurchased_From, SI_sProvider_UniqueID, SI_bHotel_Pre_Loaded, SI_sSiteName FROM dbo.SI_Provider_Profile WITH (HOLDLOCK TABLOCKX)')
GO
-- Rename the primary key constraint or unique key In SQL Server constraints such as primary keys or foreign keys are objects in their own right, even though they are dependent upon the "containing" table.
EXEC sp_rename 'dbo.SI_Provider_Profile.PK_SI_Provider_Profile', 'PK_SI_Provider_Profile_Old';
GO
-- backup old table in case of
EXECUTE sp_rename N'dbo.SI_Provider_Profile', N'SI_Provider_Profile_Old', 'OBJECT'
GO
EXECUTE sp_rename N'dbo.Tmp_SI_Provider_Profile', N'SI_Provider_Profile', 'OBJECT'
GO
ALTER TABLE dbo.SI_Provider_Profile ADD CONSTRAINT
PK_SI_Provider_Profile PRIMARY KEY NONCLUSTERED
(
SI_lProvider_Profile_ID
) WITH( PAD_INDEX = OFF, FILLFACTOR = 90, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
GO
COMMIT TRANSACTION
ALTER TABLE TABLE_NAME MODIFY (COLUMN_NAME DROP IDENTITY);
I would like to write a single SQL command to drop multiple columns from a single table in one ALTER TABLE statement.
From MSDN's ALTER TABLE documentation...
DROP { [CONSTRAINT] constraint_name | COLUMN column_name }
Specifies that constraint_name or column_name is removed from the table. DROP COLUMN is not allowed if the compatibility level is 65 or earlier. Multiple columns and constraints can be listed.
It says that mutliple columns can be listed in the the statement but the syntax doesn't show an optional comma or anything that would even hint at the syntax.
How should I write my SQL to drop multiple columns in one statement (if possible)?
For SQL Server:
ALTER TABLE TableName
DROP COLUMN Column1, Column2;
The syntax is
DROP { [ CONSTRAINT ] constraint_name | COLUMN column } [ ,...n ]
For MySQL:
ALTER TABLE TableName
DROP COLUMN Column1,
DROP COLUMN Column2;
or like this1:
ALTER TABLE TableName
DROP Column1,
DROP Column2;
1 The word COLUMN is optional and can be omitted, except for RENAME COLUMN (to distinguish a column-renaming operation from the RENAME table-renaming operation). More info here.
Summarizing
Oracle:
ALTER TABLE table_name DROP (column_name1, column_name2);
MS SQL Server:
ALTER TABLE table_name DROP COLUMN column_name1, column_name2
MySQL:
ALTER TABLE table_name DROP column_name1, DROP column_name2;
PostgreSQL
ALTER TABLE table_name DROP COLUMN column_name1, DROP COLUMN column_name2;
Be aware
DROP COLUMN does not physically remove the data for some DBMS. E.g. for MS SQL. For fixed length types (int, numeric, float, datetime, uniqueidentifier etc) the space is consumed even for records added after the columns were dropped. To get rid of the wasted space do ALTER TABLE ... REBUILD.
create table test (a int, b int , c int, d int);
alter table test drop column b, d;
Be aware that DROP COLUMN does not physically remove the data, and for fixed length types (int, numeric, float, datetime, uniqueidentifier etc) the space is consumed even for records added after the columns were dropped. To get rid of the wasted space do ALTER TABLE ... REBUILD.
This may be late, but sharing it for the new users visiting this question.
To drop multiple columns actual syntax is
alter table tablename drop column col1, drop column col2 , drop column col3 ....
So for every column you need to specify "drop column" in Mysql 5.0.45.
The Syntax as specified by Microsoft for the dropping a column part of an ALTER statement is this
DROP
{
[ CONSTRAINT ]
{
constraint_name
[ WITH
( <drop_clustered_constraint_option> [ ,...n ] )
]
} [ ,...n ]
| COLUMN
{
column_name
} [ ,...n ]
} [ ,...n ]
Notice that the [,...n] appears after both the column name and at the end of the whole drop clause. What this means is that there are two ways to delete multiple columns. You can either do this:
ALTER TABLE TableName
DROP COLUMN Column1, Column2, Column3
or this
ALTER TABLE TableName
DROP
COLUMN Column1,
COLUMN Column2,
COLUMN Column3
This second syntax is useful if you want to combine the drop of a column with dropping a constraint:
ALTER TBALE TableName
DROP
CONSTRAINT DF_TableName_Column1,
COLUMN Column1;
When dropping columns SQL Sever does not reclaim the space taken up by the columns dropped. For data types that are stored inline in the rows (int for example) it may even take up space on the new rows added after the alter statement. To get around this you need to create a clustered index on the table or rebuild the clustered index if it already has one. Rebuilding the index can be done with a REBUILD command after modifying the table. But be warned this can be slow on very big tables. For example:
ALTER TABLE Test
REBUILD;
For MySQL (ver 5.6), you cannot do multiple column drop with one single drop-statement but rather multiple drop-statements:
mysql> alter table test2 drop column (c1,c2,c3);
ERROR 1064 (42000): You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near '(c1,c2,c3)' at line 1
mysql> alter table test2 drop column c1,c2,c3;
ERROR 1064 (42000): You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'c2,c3' at line 1
mysql> alter table test2 drop column c1, drop column c2, drop c3;
Query OK, 0 rows affected (0.64 sec)
Records: 0 Duplicates: 0 Warnings: 0
mysql>
BTW, drop <col_name> is shorthanded for drop column <col_name> as you can see from drop c3 above.
If it is just single column to delete the below syntax works
ALTER TABLE tablename DROP COLUMN column1;
For deleting multiple columns, using the DROP COLUMN doesnot work, the below syntax works
ALTER TABLE tablename DROP (column1, column2, column3......);
Generic:
ALTER TABLE table_name
DROP COLUMN column1,column2,column3;
E.g:
ALTER TABLE Student
DROP COLUMN Name, Number, City;
alter table tablename drop (column1, column2, column3......);
for postgis is
alter table table01 drop columns col1, drop col2
this query will alter the multiple column test it.
create table test(a int,B int,C int);
alter table test drop(a,B);
ALTER table table_name Drop column column1, Drop column column2,Drop column column3;
for MySQL DB.
Or you can add some column while altering in the same line:
ALTER table table_name Drop column column1, ADD column column2 AFTER column7;