Data historian queries - sql-server

Question

I have a table that contains data for electric motors the format is:
DATE(DateTime) | TagName(VarChar(50) | Val(Float) |
2009-11-03 17:44:13.000 | Motor_1 | 123.45
2009-11-04 17:44:13.000 | Motor_1 | 124.45
2009-11-05 17:44:13.000 | Motor_1 | 125.45
2009-11-03 17:44:13.000 | Motor_2 | 223.45
2009-11-04 17:44:13.000 | Motor_2 | 224.45
Data for each motor is inserted daily, so there would be 31 Motor_1s and 31
Motor_2s etc. We do this so we can trend it on our control system displays.
I am using views to extract last months max val and last months min val.
Same for this months data. Then I join the two and calculate the difference
to get the actual run hours for that month. The "Val" is a nonresetable
Accumulation from a PLC(Controller). This is my query for Last months Max
Value:
SELECT TagName, Val AS Hours
FROM dbo.All_Data_From_Last_Mon AS cur
WHERE (NOT EXISTS
(SELECT TagName, Val
FROM dbo.All_Data_From_Last_Mon AS high
WHERE (TagName = cur.TagName) AND (Val > cur.Val)))
This is my query for Last months Max
Value:
SELECT TagName, Val AS Hours
FROM dbo.All_Data_From_Last_Mon AS cur
WHERE (NOT EXISTS
(SELECT TagName, Val
FROM dbo.All_Data_From_Last_Mon AS high
WHERE (TagName = cur.TagName) AND (Val < cur.Val)))
This is the query that calculates the difference and runs a bit slow:
SELECT dbo.Motors_Last_Mon_Max.TagName, STR(dbo.Motors_Last_Mon_Max.Hours - dbo.Motors_Last_Mon_Min.Hours, 12, 2) AS Hours
FROM dbo.Motors_Last_Mon_Min RIGHT OUTER JOIN
dbo.Motors_Last_Mon_Max ON dbo.Motors_Last_Mon_Min.TagName = dbo.Motors_Last_Mon_Max.TagName
I know there is a better way. Ultimately I just need last months total and this months total. Any help would be appreciated.
Thanks in advance

Answer 1

First two queries can be handled as one. Something like:
SELECT TagName, MAX(Val) AS MaxVal, MIN(Val) AS MinVal
FROM dbo.All_Data_From_Last_Mon
GROUP BY TagName
-- ORDER BY TagName (optionally)
I now see that these queries are SQL views, used for the third query... and I can see why this would be slow ;-)
The following reproduces the logic, but without the views, and this should allow SQL to optimize quite a bit. At any rate it provides more clarity as to what is being done...
Please "give it a spin".
SELECT DISTINCT Mx.TagName, STR(Mx.Hours - Mn.Hours, 12, 2) AS Hours
FROM dbo.All_Data_From_Last_Mon Mx
RIGHT OUTER JOIN dbo.All_Data_From_Last_Mon Mn ON Mx.TagName = Mn.TagName
AND dbo.All_Data_From_Last_Mon -- Cut the cross product a bit; may not be necessary
WHERE
NOT EXISTS (SELECT * FROM dbo.All_Data_From_Last_Mon Mx1
WHERE Mx1.TagName = Mx.TagName AND Mx1.Hours > Mx.Hours)
AND NOT EXISTS (SELECT * FROM dbo.All_Data_From_Last_Mon Mn1
WHERE Mn1.TagName = Mn.TagName AND Mn1.Hours < Mx.Hours)
Notes:
- notice the DISTINCT in SELECT statement. That is to avoid dup lines in he case there would be several days that show the Maximun (or minumum) Hours value for that month.
- the extra condition on the join is aimed at avoiding a full 31 * 31 cross product, but the conditions that truly bring it to a single line (or several in case of dups) are the NON EXISTS predicates that follow.
- A TagName+Hours index, if not readily present would greatly help.
==> I'd be interested in feedback on this query performance, as run with actual data.