There's two parts to this. First is getting it into SQL, which you can do like this: (props to whoever provided this sample. it didn't say)

DECLARE @FileContents VARCHAR(MAX)

SELECT @FileContents=BulkColumn

FROM...

There's a textbox in multiline mode bound to your data.

Ding Ding Ding! We HAVE a winner! And I'm not just saying it 'cause our names sound alike.

you should change the key.

[UtcTime] [datetime] NOT NULL, -- clustered not unique

--not part of index

[MachineRegisterId] [int] NOT NULL,

[Value] [float] NOT NULL

and add an unique index on utctime and machineregisterid, so...

Look, I apologize in advance but,

Where are you going with this?

Thanks, Ladies and Gentlemen, I'll be here all week.

GilaMonster (7/22/2014)

---

My question still stands. Why are you retrieving all 4 million rows and sending them to an application. Is there no filtering or aggregating that can be done in...

Could the problem here be the 12 column compound key? We got no table definition to look at, but that seems awful wide. One way to test that would be...

I believe in trying to help every programmer be the best they can possibly be - otherwise when I crush them utterly it will be meaningless 😀

Yep, I understand what you're saying. Really, I don't think the business requirement was well thought out (who breaks ties) but I've tried to explain these things to stakeholders before...

You're almost there. You just have to add one more line to your query. Instead of

SELECT @SQLStatement = @SQLStatement + ',' + c.name +' '+ CASE

WHEN st.name LIKE '%CHAR%'

THEN...

OUTER APPLY instead of CROSS APPLY.

if you wanted to force that business requirement to work, maybe you could insert the original request with a priority of say, 2,000,000,000 and with all future requests decrement that...

Any time you have consistent date math it's much faster to go

select * from

(select convert(date,getdate()-1) d1,convert(date,getdate()) d2) a

outer apply

(Select...

select * from

(

SELECT [columns], row_number() over (partition by c.id order by s.start desc) rn

FROM Client c

INNER JOIN Services s ON

c.SomeColumn = s.SomeColumn

WHERE s.START > GETDATE()-80

) a where rn=1

You could try updating your index to be like this

CREATE INDEX IDX_Abstract ON #Diag_Codes(AbstractID,EpisodeNumber,OrderOfDiagnosisInEpisode)

INCLUDE (rowupdatedatetime,DiagnosisIcd)

That way it will be completely covered and won't have to do key lookups.