Month: October 2014

Maintenance Plans – I don’t think they are doing it right

Are they rebuilding all their indexes and then updating stats at 20%?
Or, perhaps, rebuilding all indexes where fragmentation is greater than 50%, reorganizing all indexes where fragmentation is greater than 20% and not updating any statistics at all?
Or, my personal favorite, autogrow in 1MB increments followed by a weekly shrink. OH YEAH!”

Objectively point out where the maintenance plan does not line up with this link:

Let the data do the talking. If they still circular file your recommendations, then you have to do a risk assessment on whether it is worth having the DBAs hate you when you go over their heads.

I don't know how to answer that question


When there needs to be a change to the ‘standard configuration’

Screaming fit

What do you do when a client complains about slow running queries, but when you suggest changing Maximum Degree of Parallelism to something other than “1”, you are told, “That is our standard configuration. Company policy.”
“Why?” you ask.
“To avoid CPU utilization pressure on other queries,” they reply.

If you want to be the hero (admit it, its okay. we all want to be the one who walks in, hits CAPS LOCK and walks away) then you are going to have to justify the change with hard data:

Create a VM with the ‘company standard’ settings and time the queries.
Change the settings and time the queries.
If there is a significant difference, stick the results into an Excel spreadsheet and make a pretty graph.
Print 10 copies and distribute. No soft skills required. 😉

Note: if the database is too large to restore a backup onto a VM, then create a clone of the database minus the actual data: