Reputation: 19718
I'm currently working on a data migration project and for performance-related issues, I want to predefine a set of identities rather than letting the tables generate them.
I found it's not easy to add the identity
property to a column, so I want to use IDENTITY_INSERT ON
statement.
My question is: would this disable updates to the table's identity table (which is impacting performance), or do I need to truly remove the identity
property of the column(s)?
Upvotes: 3
Views: 13122
Reputation: 2380
Something to note about Identity columns with SET IDENTITY_INSERT ON.
Just checked on SQL 2012 you can't insert using the built in auto identity if you turn the option on.
Quick test below...
BEGIN TRY DROP TABLE #T END TRY BEGIN CATCH END CATCH;
CREATE TABLE #T (id int IDENTITY, Name varchar(50));
INSERT INTO #T (Name) VALUES ('Darren'); -- built in Identity format
SET IDENTITY_INSERT #T ON;
INSERT INTO #T (id, Name) VALUES (5, 'Jerry'); -- explicit format of identity
INSERT INTO #T (Name) VALUES ('Thomas'); -- TRY to use built in format
SET IDENTITY_INSERT #T OFF;
SELECT * FROM #T;
results with ...
(1 row(s) affected)
(1 row(s) affected)
Msg 545, Level 16, State 1, Line 11
Explicit value must be specified for identity column in table '#T__________________________________________________________________________________________________________________000000C34998' either when IDENTITY_INSERT is set to ON or when a replication user is inserting into a NOT FOR REPLICATION identity column.
(2 row(s) affected)
Upvotes: 0
Reputation: 7376
It's very common for data migration scripts to have something like:
SET IDENTITY_INSERT [MyTable] ON
INSERT INTO [MyTable] ...
INSERT INTO [MyTable] ...
INSERT INTO [MyTable] ...
...
SET IDENTITY_INSERT [MyTable] OFF
While enabled, the field will not auto-increment for other inserts.
IDENTITY_INSERT has session scope, so only your session will be able to insert to the identity row explicitly. AND only one table in a session can have IDENTITY_INSERT ON at a time.
So what about performance? I don't actually have an answer to you question, but I have some code that should give you an answer. It's a modified version of something I found here:
/* Create a table with an identity value */
CREATE TABLE test_table
(
auto_id INT IDENTITY(1, 1),
somedata VARCHAR(50)
)
GO
/* Insert 10 sample rows */
INSERT INTO test_table
SELECT 'x'
GO 10
/* Get the current identity value (10) */
SELECT Ident_current('test_table') AS IdentityValueAfterTenInserts
GO
/* Disable the identity column, insert a row, enable the identity column. */
SET identity_insert test_table ON
INSERT INTO test_table(auto_id, somedata)
SELECT 50, 'x'
SET identity_insert test_table OFF
GO
/* Get the current identity value (50) */
SELECT Ident_current('test_table') AS IdentityValueAfterIdentityInsertWithIdentityEnabled
GO
/* Disable the identity column, insert a row, check the value, then enable the identity column. */
SET identity_insert test_table ON
INSERT INTO test_table(auto_id, somedata)
SELECT 100, 'x'
/*
Get the current identity value (?)
If the value is 50, then the identity column is only recalculated when a call is made to:
SET identity_insert test_table OFF
Else if the value is 100, then the identity column is recalculated constantly and your
performance problems remain.
*/
SELECT Ident_current('test_table') AS IdentityValueAfterIdentityInsertWithIdentityDisabled
SET identity_insert test_table OFF
GO
/* Get the current identity value (100) */
SELECT Ident_current('test_table') AS IdentityValueAfterIdentityInsertWithIdentityEnabled
GO
DROP TABLE test_table
I don't have a SQL SERVER handy to run this on, so let me know how it goes. Hope it helps.
Upvotes: 10