How to Capture Baselines with sp BlitzFirst
Vložit
- čas přidán 14. 02. 2019
- Everybody tells you to capture baselines - but what exactly are you supposed to capture? Which Perfmon counters matter? How do you track which data files are growing, and which ones are slow? How can you track which queries are using the most resources?
- Věda a technologie
It is unbelievable . How this much powerful queries given away for free. Hat's off to you sir.
Mr. Brent Ozar you are a funny genius that brings simple truth to a highly complex disoriented world. So refreshing your clear reasoning and sensibility. Kudos to the solutions you disseminate masterfully. You are a legend in your own time. It's my honor to reap what you sow. Cheers!
Hahaha, thanks!
Thanks Brent! it really is amazing. I’ve tested on my SQL Server and I’m impressed with the outcome.
It's a bit over my head at the moment, but I can't stand the baseline challenge and look forward to figuring this out as far as analyzing the results and taking action. I've probably been totally wrong in my baseline approach. I'll run performance test projects that simulate a load of users, while going through detailed parts of the application myself and capture all that traffic in a replay trace. Note, not a distributed replay trace, since I'm still too big of a nitwit to set that up. I then use that traffic against toggling compat to get a general idea of how we'll work with the new CE (I really hoped it'd be better after reading your article on it eating sql for breakfast), but it really choked bad when indexes were missing, even after enabling TF 4119. I can't really get the stats off production though, so I have to figure out how to fake it well.
Your content has been extremely helpful since I've operated as the dba. I last used your query tuning video against some pretty complex cte driven sql and was able to easily poke holes in things like union usage (dropped my reads by a third), implicit conversion (knew views needed count_big, but didn't know about ctes!) and saw some general benefits from trial and error like swapping group bys for distincts and getting prefiltering. Pulling the IO stats into something generally friendly to look-at (I always squinted like a fool, which made me hate using them) was fantastic. I was able to allow the team to remove a performance filter that skewed results by making the query faster than it was with the filter. Anyway I'm sure you hear this stuff all the time.
Thanks again, appreciate you donating your time for these though. Sorry for the novel.
Thanks, glad you like our work!
Such amazing info thanks a lot Will apply ASAP keep doing it! Thanks a lot Brent! U r crazy dude XD
Brent - great information, something that is needed where I currently work. Do you have a script that will build out the tables and views in the DBAtools database? I can look at the result sets and do it myself, but figured if it was already out there somewhere it make my work a bit easier. Thanks!
Gill - Thanks! It's done automatically by sp_BlitzFirst when you run it.
Thanks - I'll have to look through the code to see where that happens - it didn't create anything for me, just the result sets
@@gillrowley7264 That means you didn't use the parameters (OutputDatabaseName, SchemaName, TableName, etc.)
Tried them at first without success, but I got it running. Great stuff. Thanks!
Another great video, Brent! Thank you very much! A pity that Power BI is not free for private users, though :(
You're thinking of the online service. The Power BI Desktop is indeed free, and that's what the video's about.
Thanks Brenzt , how can I find a baseline value which a now normal from this to show management?
To learn about writing reports, check out my Mastering Server Tuning class.