[Video] Office Hours: Black Friday Edition, Ask Me Anything About SQL Server

Videos
0

Yesterday, I finished setting up my new Apple MacBook Pro with the new M1 Max processor and did its first live stream on my Twitch channel to see how the hardware held up. I was tickled pink – worked great!

Here’s what we covered:

  • 00:00 Introductions
  • 03:48 TheGreatFode: Hi Brent, have you ever seen cases where to low HBA queue depth setting was a problem, and increasing it improved I/O performance? BusyIsGood: Hi Brent, I’m trying to use sp_BlitzLock to get recent deadlocks but the procedure has started taking over 30 minutes to produce records. What could be causing this delay? Or is this just my life now…
  • 09:33 NewPadawan: Hey Brent, is splitting tempdb still needed on SSD? On the company I work at, we have about 40 databases with each db is about 20-40GB, on 1 SSD. We only use 1 tempdb (on another SSD) so far, on sqlserver 2008 r2.
  • 10:39 BusyIsGood: Hi Brent, I’m trying to use sp_BlitzLock to get recent deadlocks but the procedure has started taking over 30 minutes to produce records. What could be causing this delay?
  • 12:15 LookingAtProcsAllDay: Hi Brent I’ve seen many different upsert patterns (some of them considered anti) in TSQL. Everything from: If (not) exists / merge / @@rowcount / updlock / serializable / list goes on… What is your “go to UPSERT pattern” in TSQL and why?
  • 15:20 sol: Howdy, sir, tasty be your beverage! Earlier this century twas like “MaxDOP at 50? – Brent wrote this… – Who’s Brent?!” and now it’s more like “Why..? – Thus Brent Hast Written – Ooh, so wise!”. By giving more and more to the community, are you afraid of becoming “the default”?
  • 18:04 i_use_lowercase_for_select: Hi Brent, My friend who is also my customer has a 100GB database and a terrible application which uses ‘select * from [table]’ for every query. Performance is terrible and now he wants to use ephemeral disks for data and logfiles. Can you help me expess how bad of an idea that is
  • 20:25 Lostindb: Hi Brent, I used to put mdf and ldf on separate physical disk location. But it is still needed when we use SSD? Or is it safe to put mdf and mdf in same SSD? Thanks!
  • 21:26 RelationalDBA: Hey Brent, love your office hours videos. If you were a Production DBA, what would your approach be to monitor AlwaysON AG? My friend’s office decided to have HA on the DBs and he wants to be prepared.
  • 24:22 Accidental DBA: Hi Brent, sometimes when we run sp_blitzfirst with no arguments, it runs for several minutes (15 minutes). Is this something to be concerned about? Looking at messages tab, it says “Beginning investigatory queries”. SQL2014
  • 25:28 Stuart: Hi, on SQL2016 I have a query using outer-join to empty columnstore index table to benefit from batch mode; on 2019 (compat150) if I remove the outer join to columnstore table it gets a very different plan and no batch mode so seems this is still useful on 2019 – any thoughts?
  • 27:50 Stuart: Hi Brent, I recently created a high-value missing index on a single (computed) column, according to sp_BlitzIndex. When I check the index usage stats DMV days later though it shows it’s not been used at all, any thoughts on why that might be?
  • 28:54 Ivan: Is it necessary to rebuild index which had page count more then 500 above & fragmentation more then 50% daily or weekly once should be fine ?
  • 31:09 Wrap-up

Wanna get alerted whenever I go live? Subscribe to my Twitch channel. See you there!

Previous Post
[Video] Office Hours Speed Round: 17 Answers in 11 Minutes
Next Post
Today is the last day to buy a Live Class Season Pass, EVER.

Leave a Reply

Your email address will not be published.

Fill out this field
Fill out this field
Please enter a valid email address.

Menu