As part of SQL Response V2 we’ll be collecting all sorts of data that allows the user to see what is happening on SQL instances monitored by the product. However, obviously, we have a delicate balancing act to perform – we want to maximise the information we obtain whilst minimising the impact on the monitored machine.
As most DBAs know, running a SQL Trace on a box will provide a great deal of information about what’s happening on the server – what statements were executing immediately before and after the alert for example. The downside of course is the impact – a few percent on CPU at least, not good on a production server.
Does this stop you using SQL Trace? Have you ever found it useful, even vital, analysing server problems or do you use other sources?