Nov 16 2009
It’s been a week now since the 2009 PASS Community Summit ended. The last of us filed out of Seattle on Monday and here it is, Monday all over again. So did you implement anything you learned at the Summit yet? All those fresh ideas in your head are starting to get swept out by those clock hands. If you don’t implement soon, then it’s all for naught (at least until those Summit DVDs show up in the mail.)
Here is my challenge to you: if you know you’ll not have an opportunity to implement at least three new processes or concepts from this year’s Summit, take the time to review your notes and flesh out the details to a point that you can pick them up and hit the ground running in a week, a month, or 4 months when you do have time. I know that when I was reviewing my notes from Maciej Pilecki’s well-done Dude, Where’s My Memory session, that I was already having difficulties deciphering my frantic scribblings. (PASS, please do a better job of ensuring all the speaker’s session slides are available prior to the session.) (Oh, and speakers, please do a better job of getting your slide decks and scripts to PASS HQ prior to the start of your session.)
What have I implemented in the last week? Well, I’d already been collecting wait stats from the sys.dm_os_wait_stats DMV and simultaneously was collecting and persisting the concurrent results from each of the following DMVs:
However I was not collecting file wait stats (courtesy of sys.dm_io_virtual_file_stats). After attending Andrew Kelly’s session on collecting File and Wait Statistics, I knew that I needed to augment my current processes to do just that. I also liked his reporting mechanism and altered my reporting to match his supplied scripts. I did make some changes to his code though. I thought it lacked the key metric of file reads/sec and file writes/sec so I created that calculated value from the supplied columns from the sys.dm_io_virtual_files_stats DMV. I also assign an increment_id value to each collection cycle for the five DMVs I collect data from for analysis of file I/O and wait stats. Therefore I needed to alter his table creation script, collection, and reporting scripts to accomodate my usage patterns. I found no issues with his supplied scripts from the PASS Summit 2009 download site. They were functionally accurate; they simply did not match my standards for the environment I support out-of-the-box.
I still intend to listen to Glen Berry’s Dr. DMV session to see if I can glean additional knowledge on the DMVs he covered. I also plan on using what I learned from Maciej’s session to more-fully assess instances that may have improperly-configured memory settings.
What have you implemented from the PASS Summit 2009? Did you find yourself using the code available without tweaks? If not, what did you have to do in order to get the code to function in your environment? What do you still intend to implement? I ask these questions of Jeremiah Peschka (twitter | web), Andy Leonard (twitter | web), and Jen and Sean McCown (twitter | web). I ask that they tag 2-3 more individuals with their response and link back to this original post. The goal is to show both first-time Summit Attendees and Veterans alike (and their Managers/Directors) that there is value in what you bring back from the Summit in that wetware you call a brain.