We are running the portal in a web farm environment and are using  Broadcast polling.  Even though the scheduled job for Broadcast polling is scheduled to run every minute, we see in the SchedulehIstory table that sometimes the job runs 18-20 times in a minute on a server.  We have seen this consistenty in all our environments.  Has anyone run into this problem and is there an issue with the scheduler in the Web Farm environment?  We are also getting a lot of deadlocks and timeouts in Evenbt log.  We think that this job along with some others migth be responsible for the timeouts.  We have also seen that the following SQL statement in stored procedure [PurgeBroadcasts] takes a long time (upto 4 seconds) to execute.

 
DELETE FROM dbo.Broadcasts
            WHERE BroadcastDate <
                  (SELECT Min(LastActivityDate) FROM dbo.Servers WHERE ServerID<>dbo.Broadcasts.SourceServerID)
                 
I changed the SQL statement to modify the sub query and it runs almost instantly.  Changes are indicated in Red below. 

DELETE FROM dbo.Broadcasts

            WHERE BroadcastDate <

(SELECT Min(LastActivityDate) FROM dbo.Servers WHERE ServerID not in (select distinct SourceServerID from dbo.Broadcasts))

Does anyone see any problems with the modified query? 

It will be great if others can share what kind of database optimizations or other tuning have they done to tune optimize the performance for DNN?

Thanks in advance,

Chhavi