Hey, I'm still around. I can't say that I know too much about this process as I got it working quite a while ago. From what I can see from...

Jeff Moden (2/25/2012)

---

I also made the assumption that the meter readings will actually be unique by the datetime of SampleDate and MethodID (and they really should be or you have...

Thanks for the insight everyone. And thanks again Jeff for going over and above to help find those roadblock cross joins. I will try the composite index and see how...

Statistics on the base table are up to date, that didn't fix the issue.

The view is just so I don't have to type that join all of the time....

Jeff Moden (11/26/2011)

---

loki1049 (11/25/2011)

---

So we I have tried to split the string, which will be in the form of "X.XXEXX,X.XXEXX,X.XXEXX," but using the split function like say the DelimitedSplit8k will...

So looking at this more, it seems we have to store it as a comma delimited string in an NVARCHAR(MAX) column due to the strings length*. Anyhow, its a similar...

Ah yes, ok I can work with this. Thanks for the example. I will be getting back at this tomorrow.

EDIT:

Jeff Moden (11/21/2011)

---

Heh... almost forgot. Now you really know why...

Jeff Moden (11/21/2011)

---

Just to clarify... you said the range would be from 0 to 50000 counts (I think I know what that means but you might want to clarify). ...

Jeff Moden (11/21/2011)

---

I don't know what kind of output your spectrometers can cough out but, if they can use the kind of "concatenated binary" format I'm thinking of, then SQL...

Jeff Moden (11/18/2011)

---

For a spectrometer, you would expect all numeric values. Rather than bother with XML, why not store it all as "fixed length" fields within a fixed length...

That's just the thing, we are not really...

Thanks for the very informative post. Sorry I have not participated but I was away for the weekend. As Jeff pointed out, yes the spectral data is 1024 randomized (due...

SQL server limits the number of columns in a table and view to 1024. So the wide solution is not acceptable as it stands because 1024 columns plus 1 sample...

Lowell (11/16/2011)

---

hybrid of the wide table idea?

do those 1024 columns logically break into groups that could be, say 4 or 5 other tables? (measurement types?)then you have , dunno, 5...

mtassin (4/18/2011)

---

That way the quick FMTONLY on statement will just get a field list back.

My problems with FMTONLY have involved temp tables not getting created in that case......