Please post the complete code of the procedure called VTLSDBLoadTest.dbo.SaveTrackingRecord

Hmm you are right.

Try this instead:

;with cte as (

select

t1.*,

-- get the id of the previous non-zero row

(select top 1 id from ##temp t2 where t2.did=t1.did and t2.id <...

The autogrow might cause file fragmentation. To see the current status of the database files you could use contig.exe

http://technet.microsoft.com/en-us/sysinternals/bb897428.aspx

contig.exe can also be used to defragment a single file. That is...

It will probably not make much difference regarding performance.

I expect the four int columns to be slightly faster.

After reading your other post a bit more carefully I now think that what you really want is something like this:

CREATE TABLE ##temp

(

ID INT IDENTITY(1,1),

DID INT,

Value INT,

Date DateTime

)

INSERT INTO ##temp...

Your description of the requirements is really hard to understand, but the way I interpret you this should work:

select * -- replace with delete to actually delete

from ##temp t1

where

value...

Glad to hear that it works.

From your posted results it should be possible to improve the performance even more.

You only have 1800 grid squares but the most dense squares have...

Some notes:

Make sure you have a unique clustered index on IPLocationDB_Location(location)

Make sure you have a unique nonclustered index on IPLocationDB_Region(CountryCode, RegionCode) INCLUDE(RegionName)

You should be able to remove the DISTINCT from...

Your query looks ok to me. Is there still any problem ?

Of course you still need to test the code so the result is what you expect.

This should work just fine:

-- Add a non-clustered covering index to help query performance

create index ix1 on dbo.EffectiveItems(grid, low)

include(ItemID, high, [begin], [end], Val1, Val2, Val3, Val4)

select ItemID, Val1, Val2,...

But the size of the backup is not affected by the unused part of the transaction log.

If only 1% is used, only 1% will be backed up.

One thing to be...

Stefan 😉

If I understand you correctly, the columns with the business keys in the fact table will ALLWAYS slow performance, even though they will never be included in...

I think the following will give much better performance:

-- Create an index to speed up lookup by ip

CREATE INDEX IX1 ON [IPLocationDB_Organization](startIP) INCLUDE(endIP, organization)

-- Execute reformulated query

SELECT

CompanyVistedProfile.VisitedDate,

ISNULL(

(SELECT organization

FROM

(SELECT...

Just curious: Did you test my last suggestion ? How did it work for you ?

Your queries against this fact table will most likely result in scans of large ranges of data. Most likely data will have to be read from disk.

If that is the...