[Video] Office Hours: Ask Me Anything About Microsoft Databases
I’m back at home in Vegas, taking your top-voted questions from https://pollgab.com/room/brento on a hillside enjoying the fall desert weather.
Here’s what we covered:
- 00:00 Start
- 03:10 Bruce: Howdy! When, if ever, would you recommend implementing a SQL Server Central Management Server?
- 04:09 Impostor Syndrome: Are multi-TB databases really all that rare? I have several that are over 10 TB, but I don’t think of myself as being in a top 1% company.
- 05:23 LikeHeardingCats: Hey Brent. You’ve mentioned in the past, I believe, that you deal with Imposter Syndrome. Being the SQL Guru that you are, do you still have doubts when it comes to SQL Server? If so, how do you manage those thoughts?
- 07:29 Stefano: Hi Brent, I make extensive use of temporary tables (
- #table) for complex reporting or data extraction queries, particularly for preprocessing data from linked servers (yes, they exist…). My tempdb is fast and capacious. Are there any drawbacks to this approach?
- 10:31 Dopinder: What’s your opinion of the various AI engines ability to optimize TSQL queries and sprocs? Which one is best?
- 12:02 Culloden: Hey Brent, Have you participated in any recent DBA interviews for your clients? If so, have you noticed any trends in what skillset employers are seeking?
- 13:45 Parameter Sensitive Boyfriend: Terminology question: if I have a query with OPTION(RECOMPILE) that has massive variation in performance based on what arguments are passed in, is it considered parameter sensitive?
- 14:47 MyTeaGotCold: People tell me that 32 GB of RAM is enough for a high-end gaming PC. I’m looking to build one, but my DBA instincts say that anything less than 100 GB is unacceptable. How do you resolve this conflict in your machines?
8 Comments. Leave new
Please do an office hours video from the cruise if you can, that sounds like an absolute riot!
So, this gets a little tricky: not everyone is out of the closet, and it would be rough for someone’s reputation if they were spotted by someone else.
I wonder, that no carmaniac grilled you for putting a metal can direct on the car paint (possible scratches). But maybe the pink color saved you 🙂
PS: regarding the #temp_table quesion: you argued that it will consume RAM, but without a #temp it would use the same amount of RAM too (for the same task) and maybe even more, because SQL Server tends to do too much work on rows that are filtered out later. Or it thinks that you get only 1% of the real amount of rows back when joining on an UNID because it simply doesn’t know that every row in tbl_1 has a macht in tbl_2 and you have no Foreign Key (e.g. because it is a Columnstore Index table)
Hahaha, the car is wrapped (with two layers, actually – pink, then a clear metallic layer over that.) It’s originally a dark brown that has a lot of red to it. Looks black when it’s inside, but outside it varies a lot from brown to red. Rolls does a heck of a job with their paint colors.
I’m not sure I agree about the temp table question. In the Mastering Query Tuning class, I actually demo cases where working with a temp table is worse if you don’t do the filtering at the right step. Just as SQL Server can get the filtering in the wrong place, human beings certainly do it too.
True story: I was working with a client this week that had a multi-step reporting process that had been gradually tacked-on for years. The report proc called one proc, which called another, which called views, etc. It didn’t even take 30 seconds of digging for me to point out that the first stages of the work were involving data that simply weren’t used in the final steps – they’d just kept copy/pasting the code over and over, dumping an initial broad set of data into temp tables, and then filtering it differently with subsequent pieces of code.
Was the code bad? Absolutely. But generally speaking, I think most code is, hahaha.
First up, thank you for the 360 degree camera as I appreciated seeing all the doggos!
Secondly (and you and others will no doubt roast me in this), I do not nor do I see me actively using AI do to my DBA job. And it’s not for fear of it taking over the world (though with the uptick in SORA content I can see it having a massive impact for sure), it’s more that I feel as if I’ve spent the last two years really focused on education and upping my DBA role value that AI in a way cheapens that. As if it somehow invalidates all the energy I’ve put into it. Why bother learning how to properly tune a query or index if I can just feed one of these tools the info and have it spit out the answer? And when others in my organization say “Well, I put it in co-pilot and here’s what it says”, that’s a personal slap in the face to me. My knowledge and answer isn’t good enough so I’ll see what AI responds with.
Now, I can imagine your respond being “it’s another tool in the belt to choose from.” I get that. I am trying, I want to continue to be useful to our organization. However, I’m also on the back side of my career and if some point I get replaced by a “machine”, getting something new will be an uphill slog.
Ok, rant over. I still love all of your content, will continue to support and can’t wait to meet you next month in Houston!
Brian – you’re welcome! Honest question: do you use IntelliSense and the DMVs?
DMVs, yes. Intellisense, truthfully, only to allow to check that my tables, columns, etc. exist. No so much for autofilling the remaining text. We have SQL Prompt as well and I’ve completed turned it off. It’s more of hindrance than a help to me. Weird right? Why wouldn’t I want to be more productive. If I’m honest, I think it’s a pride thing. I like to show that I know what I’m doing without the assistance of a “helper” tool.
I’m weird, I know it.
Brian, I’mma be honest: you sound like a carpenter that refuses to use power tools.
I would take a few moments to think about how that might make you look to prospective employers and coworkers.