It’s the last San Diego balcony edition of Office Hours! You posted your questions at https://pollgab.com/room/brento and upvoted the ones you’d like to see.
Here’s what we covered this week:
- 00:00 Introductions
- 00:50 Steve E: Hi Brent, How are you enjoying the Porsche 944?
- 01:55 Midwest DBA: Hey Brent, The wife says my DBA oncall schedule is too intrusive to our family time. How do I switch to being a developer with no on-call? Is being a developer the only career transition I can make that would have no oncall? Will I make less or more? I’m currently at $90K
- 04:25 NotCloseEnought2RetirementToStopLearning: Brent As a prod SQL DBA since 2001, I’ve expanded my skills into Azure SQL and Cloud generally to keep relevant. It hasn’t increased my marketability as I expected. What are technologies (Postgres? Synapse?) are worth learning to stay in high demand for the next 5-7 years?
- 06:15 DBA_Mufasa: Salut Brent! What is your recommendation regarding failing over an Always-On Availability Group during Server OS patching for instance. I have “heard” it’s best practice to move the cluster resource within failover cluster manager first, before failing over the AG group in SSMS.
- 06:43 Wally: How do you deal with justified / unjustified criticism as a DBA?
- 08:18 Doug: Are there any good use cases for when stats auto update should be turned off for a given table?
- 10:45 Doug E: Hi Brent, is there a good way to know when a plan operator has a hard coded estimate?
- 11:01 Halim: What are your top two most painful mistakes as a DBA?
- 15:21 SQLing Pigs: Third Party App is sending ad hoc queries with IN statement that has varying number of items. I’m getting thousands of execution plans from this. What are my options? More important plans keep getting pushed out.
- 17:16 Tony Feuz: Brent – I know you are a fan of assigning one database to each customer instead of a multi-tenant model but if you are going down the path of multi-tenant: then is using an inline table valued function a good way to implement row level security? Are there performance issues?
- 19:54 HerbyHoover: Hi Brent, What would you say are a few core performance metrics that a Jr. SQL Server DBA should focus on better understanding?
- 20:41 Sultan: Hi Brent, should Microsoft put the TSQL debugger back in SSMS? Do you ever use the TSQL debugger in Visual Studio?
- 23:08 Wally: Which SQL server docs do you read / focus on when a new version of SQL Server is released?
- 24:25 Jim Ignatowski: Hi Brent, our top wait is PREEMPTIVE_OS_GENERICOPS (Wait time hours = 3746.1, Hours Sample = 812.6, Per Core Per Hour = 0.1, Physical SQL 2019 Enterprise with 64 cores). What can we do to reduce this wait or should we ignore it?
- 25:09 Benny Hill: Do you or your clients ever run into issues with SQL Server / TLS?
- 26:04 The F5 Masher: Hi Brent, should Microsoft enhance extended events to allow writing of session results to actual tables?
- 27:45 Latka Gravas: Hi Brent, for a dedicated backup test / checkdb SQL Server (SQL 2019, 64 cores), is it a good idea to set Maxdop to 0 to maximize CPU use? Using dbatools sql agent job to do the backup restores / checkdb’s.
- 30:18 Ralph Nader: Hi Brent, is there any safety value in running CHECKDB against TempDB and Model?
- 31:41 Jacob: Does updating stats for a table have similar free space requirements to that of rebuilding an index?
- 32:08 SQL Serenader: In one of your articles, you suggest 4 GB or 10% of RAM (whichever is LARGER) for MAX server memory. What about a server with 767 GB of RAM? Allocating 77 GB to non-SQL processes seems excessive. Could we get away with 16 GB (2%) if we don’t see memory pressure or paging to disk?
- 34:32 Genady D.: Hi Brent Recent expensive queries in Activity Monitor miss (by my opinion) link to user/host who run it. How do you deal with this problem? Which way except profiler the best?
- 35:50 Recap and goodbye to the balcony
