I would primarily note that CHECKSUM() returns a 4 byte CRC-32 value, and therefore has a very small number of possible values; a four and a half billion row table...

Having the page size be a fixed 8KB for over ten years is, in and of itself, subject to serious re-thought; are all the reasons it was chosen then still...

sgambale (2/10/2010)

---

tracing doesn't do it because I need to constantly count # of queries of certain users, not for one day, but always.

It's used to report on trends of #...

An excellent comment, SQLBOT.

Note that grouping by the TextData column is very easy if you first convert it to a VARCHAR column, and then spend the time to 'normalize' that...

Perhaps you could use a SQL Server Profiler trace that saves to a table, with minimal fields (perhaps TextData only, though I'd say add in Reads as well) and appropriate...

Joe Johnson-482549 (2/5/2010)

---

Did you read the article about encrypted drives, that...

Anyone's toaster is that which they have sufficiently relaxed requirements about in comparison with the popular products on the market that the popular products don't have any differentiation that is...

A) Set up a job that records tempDB size every minute

B) Set up SQL Profiler to record SQL:BatchComplete and RPC:Completed

The two together will give you some answers you asked for.

Another...

Heaps _as such_ cannot be defragged... but creating a clustered index and then dropping it does actually leave behind a defragged heap. This technique has, in one case I...

Find all Untrusted but Enabled constraints

Estimate the amount of time each will take to test for existing rows in violation of the constraint (DBCC CHECKCONSTRAINTS)

Determine if that will go past...

So, I believe you have something like:

A) Truncate

B) Insert

C) Update

D) More Insert

E) More Update

The most important thing is to _try things_ and benchmark how long they take, and look at...

Did you change the RAID stripe size as well and test "aligned" vs. "unaligned" at a couple different RAID stripe sizes?

Were test results consistent from one run to the next...

I'm also very interested in seeing results like this.

Was there other activity happening when the SQLIO tests were run? Multiple LUNs per RAIDGROUP, or any other spindle-sharing, can hide...

An interesting query.

I would note that if one is looking for botched foreign keys, one must look for both disabled and enabled but untrusted foreign keys and note them.

I'd also...

Jeff Moden (11/20/2009)

---

I believe that both disk I/O and time to transport to display (and any wait times the two may have) are also included in duration... as both a...