There is also a need to know that max time of the run (in the last day).  And you can't use max directly in the last query because all those...

No I mean in the code...  So that when you (or someone else) comes back 2 years from now and read that code, that you know what is happening.  Let's...

Doesn't appear so.

That seems to be the best solution so far.

I would suggest you document this code as for one have never seen anything remotely like this!

Ya I agree for the simplicity, but I just ran a perf comparaison and my if version is faster (one less table scan).  So now you have to decide how...

I don't think it's posible in a single statement but here's an idea :

CREATE TABLE A (x bigint)

CREATE TABLE B (y bigint)

INSERT A SELECT 1 UNION SELECT 2 UNION...

Quoting myself .

I misread my results.  Looks like you need to add a distinct to this query because more than 1 job step...

There's no first or last record in a table.  You just have data kept without any particular order (well order of the clustered index... but that's all).

If you want first...

I misread my results.  Looks like you need to add a distinct to this query. More than 1 job step can have the same run_time.

Maybe you could add a condition...

Sorry for the lack of formatting but I guess you can handle that at this point.

PS.  I would probabely add the job category to this report as well :

SELECT    ...

This got double posted.

Follow this link to go to the original question.

http://www.sqlservercentral.com/forums/shwmessage.aspx?forumid=8&messageid=329199

Consider this thread closed.

We need the ddl, sample data and required output on this one!!

Well then that's just have to be slow.  Creating the index would probabely take more than than selecting the data out anyways.  And since you won't gain much speed anyways...

So you get 17 rows with 3 columns instead.  If that's not what you want then you

1 - Need to know what you want

2 - Show us cause my mind...

What do you mean ("does not care").  It's not like we chose what keywords to use when the language was written.  Or what design you created. 

In this case you...

select distinct t1.Field1, t2.Field3, t2.Field4

from table1 t1

 inner join table2 on t2.field3 = t1.field1

where t1.field1 = 'char 200' AND t2.Field3 = 'char 200'   --gives a 2nd chance to use...