HoustonFirefox (11/1/2010)

---

Statements like "leave it alone", "let it grow", "never shrink it - worse thing ever" and "buy larger drives" are fine and dandy for the weekend warrior with a...

Temp tables and table variables act differently with different scopes.

The CREATE TABLE #tmp is equivalent, in the scope of the script, to CREATE TABLE tbl_table. It's not a singular...

Sounds like parameter sniffing is the culprit, offhand. As mentioned above, the .sqlplan's for your results from the above two items would help determine that.

Also, check out this...

My personal preference is a staging or temporary table to drop the load into. The reason for this is that at any specific point of failure you don't have...

<--

Hydra... um... noone here has *any* idea why it's slow, though they might have ideas on what you should look for. But, the solution? We don't even know...

Dang Jeff, much nicer than mine. I'd forgotten that even existed.

Hopefully the answer was useless, and not the...

HelsUK (10/29/2010)

---

Yes, the TRUNCATE is always working - there are always 19972 rows in my table after running the package. The identity was continuing from 19973 (not 20102 sorry)....

Sorry for that then, the hard coded values are usually reserved for homework problems.

This looks a LOT like a QotD type of trip up. Your boss quizzing your skills...

It's a trick question. What textbook and chapter? We can probably help...

HelsUK (10/29/2010)

---

Hi,

I have a SSIS package that should truncate my staging table, reseed the identity value to 100 and then insert some data into the staging table.

The problem is that...

Alvin Ramard (10/29/2010)

---

I don't agree with your thinking Craig because the identity is not metadata that is being held by SSIS. It's resides in the database engine; well, that's...

Set zoom to visible level (75%-100%).

Database Diagram menu, copy diagram to clipboard.

MSPaint - Paste - Print.

Perfect? No. But it works.

I believe this is because SSIS won't refresh metadata at runtime. This part confuses me:

Weirdly, it did work twice - the only change I made was to add 'GO'...

With no test data you'll have to adapt this to your DDL/structure, but this is one approach:

SELECT

REVERSE( LEFT( REVERSE(Col1), CHARINDEX( '.', REVERSE(Col1) ) -1 ) ),

count(*) as NumOccur

FROM

table

GROUP BY

REVERSE( LEFT(...