Hugo Kornelis (2/16/2016)

---

Sorry, but that...

Sounds like a good way to get yet more of the dreaded:

String or binary data would be truncated.

error. MS, seriously, you can't tell us which column had the error??...

These are unlikely, but let's:

1) add brackets around the db name, just in case.

2) specify a 'dbo' schema for the table name, just in case. It's easy to accidentally...

Get rid of the dashes (-) in the date, they introduce errors (ambiguity).

Use format 'YYYYMMDD [hh:mm:ss]', which is always interpreted correctly, regardless of language/date settings.

declare @CLUSTER_CODE as nvarchar(3)

declare@START_MONTH as nvarchar(2)

declare@START_YEAR...

For statistics, I'd do what DTA said. For indexes, definitely not, all DTA index recommendations should be thoroughly reviewed before creating/implementing. But for stats, if SQL "says" it...

Sorry, don't have much time but here's a quick overview of one method to consider:

1) create another table with the new bigint data type to copy the existing data to...

I would think sys.sql_expression_dependencies would contain those references, since a table-type is in sys.objects, type = 'tt'.

ALTER TRIGGER [dbo].[BadUpdate]

ON [dbo].[SomeTable]

INSTEAD OF UPDATE

AS

SET NOCOUNT ON

IF EXISTS (

SELECT 1

FROM dbo.Foo

...

Rather than being "stuck" on tuning, why not take this as an opportunity to possibly tune the query overall and post the code? Perhaps it can be reduced below...

If you create a new column, and copy existing data to that column, you don't have to UPDATE every row in a single transaction. And you don't have to...

So the entire 460GB is in the main part of the table, none in overflow space? Often with a row that wide you'd be able to offload a decent...

Yeah, should be good, except that I'd use varchar instead of char so that you don't have trailing spaces after the /s are replaced.

Sergiy (2/9/2016)

---

ScottPletcher (2/9/2016)

---

Absurd to make a blanket claim about "every page". Increasing from a smallint to a full int requires only 4 extra bytes per row (as I understand...

That sounds like Oracle not SQL Server. SS doesn't have CTAS syntax.

Maybe code below, if you're looking for a date:

SELECT CASE WHEN PATINDEX('%[_]20[0-9][0-9][0-1][0-9][0-3][0-9][_]%', string) = 0

THEN ''

...