Ah, OK. This was my first idea to solve the problem. But being an "old style" .net programmer I thought there was a more effective built in way to do it. Stream and loop, simple enough...

I'll skip the regex and find the position fo "y = " and "/>" and grab the string between those positions.

Hi all, I have a XML files that looks like this: ... <dataset> ... <p x="-0.979248" y="-0.00141319" /> ... </dataset> ... I need to find the highest value of y. The files have 100 000 rows so it should be reasonable fast. But any suggestions are welcome.

WOW! I've tried different indexes on the large table but I never got i right. Now this baby: SELECT Intervallvarden.storhet, Intervallvarden.id_intervallvarde, Intervallvarden.varde, Intervallvarden.tid, Intervallvarden.kanal_nr, MatpunktInstrument.instrumenttyp, MatpunktInstrument.instrumentnummer FROM MatpunktInstrumentParameter INNER JOIN MatpunktInstrument ON MatpunktInstrumentParameter.id_matpunkt_instrument_parameter = MatpunktInstrument.id_matpunkt_instrument_parameter INNER JOIN Intervallvarden ON Intervallvarden.instrument_typ = MatpunktInstrument.instrumenttyp AND Intervallvarden.instrument_nr = MatpunktInstrument.instrumentnummer AND Intervallvarden.tid > MatpunktInstrumentParameter.tid_upp WHERE (MatpunktInstrumentParameter.id_matpunkt = 2575) AND (Intervallvarden.tid > '2007-10-01 00:00') AND (Intervallvarden.tid < '2007-11-30 23:00') ORDER BY intervallvarden.tid Runs in two seconds! Table "Intervallvarden" being the 30 milion rows table. Thanks allot!

OK, thanks. I'll try that out. Unfourtunally I need to join on hours as well. And in a perfect world on minutes as well.

Hi all, I have a big table with 30 million rows (and growing) which I need to join to a table containing dates. The result is a bar graph diagram. But when I join with time over 12 hours the join times out SELECT table1.time, table1.value FROM table2 INNER JOIN table1 ON table1.time > table2.time_up AND table1.time < table2.time_down ORDER BY table1.time This should result in any thing from a couple of hundreds row to several thousands. I'm doing this in a Stored Procedure. I've put indexes on table1.time. Is this join simply to large for SQL Server 2005? Or how can I optimize this search?