วีดีโอ

from what I learned, the reason why it says 13.x seconds is because it is the sum of the number of seconds for each thread running in parallel. please correct me if im wrong

I want your datalength!

Long videos are like second breakfast-there's always more than you expect!

I dont have an issue with your length. You are long enough lol seriously though who is actually complaining about 15 minute videos?

Is that “live Elvis entertainment “?

Don't shorten your videos too much or I'll have to shorten my lunch break accordingly 😁 Erik, are you ever concerned the Unicode strings take twice as much storage as non-Unicode? If you will never (well, never say never, right??) have Unicode characters, is it worth the storage overhead?

When making a dynamic SQL string using multiple NVARCHAR(MAX) variables why does it sometimes not take more than 4K characters? Did any1 else face this issue here? On stack overflow I read it's due to 8K page size. But it happens only on some complex dynamic queries and I am not able to recreate it in simple scenarios.

It’s about how rounded it is

Hi, how long this 75% discount is available, I am planning to buy this in the 1st week of November, if it is still available at that time. thank you for a generous offer!

:

Shorts from Erik. O tempora, o mores!

I like it!

Whats your go to metric when gauging query performance?

Why is there no sort operator in the Query Plan at 07:55? Of course it uses the Index on the descending reputation first, but it should still need to order by the ascending Id column - except reputation is unique (in the higher ranges), but even with statistics the SQL server can't know/guarantee this, so it would need to sort. A possible reason would, if the ID is the clustered Index and SQL server guarantees, that rows would always be implizit ordered by Reputation DESC + Id ASC, even when the Id is not specified in the CREATE INDEX (the columns of clustered Index will always be included, but I in my opinion unsorted.)

So covert! Didn't think of you as that sneaky 🤣🤣

i do like devious and mischievous things...

Very interesting, what database are you using for your testing? Is it stack exchange?

My brain is mush now watching this

We went the fancy route and wrote our own code scanner with ScriptDom that gets run as part of the build pipeline. If the developer used anything smelly, we block the PR. Lol

One of my greatest pains in life is the comment from the developer "No I am 100% certain that my code is not causing any blocking because I put NOLOCK hints on all my SELECTs. So it must be something else."

I am going to add this to my toolkit. Thanks!

Kronos Quartet Plays. Chanting circle of dudes "String to String!"

Hi Erik, I replayed the same queries on my box and I have a different result, the first query (with fetch) took 10 seconds on the first run (4.2M pages read from disk), 2 seconds the next run. The second query with the row_number function was not faster altough batch mode was used. I double checked the execution plans and they are exactly the same as yours. Am I missing something? PS I use PoorMansTSqlFormatter for my formatting 🙂, and thanks for all these unforgettable parties 🥳🥳🥳

Writing code like your first example is a sure way to have your brain file for early retirement! This will do the second example: SELECT s.string, a.FirstSymbol, b.SecondSymbol, SUBSTRING(s.string, FirstSymbol, SecondSymbol - FirstSymbol) AS mySubstring FROM #strings s CROSS APPLY(VALUES (PATINDEX('%[*!$?]%', s.string) + 1)) a(FirstSymbol) CROSS APPLY(VALUES (PATINDEX('%[*!$?]%', SUBSTRING(s.string, FirstSymbol, 100))+FirstSymbol-1)) b(SecondSymbol) ;

Erik “Stimming” Darling

You may need that diaper if you continue to have problems with your colons 🤣

Bro, I'm just figuring out TVFs in general. I seem to be in more of a db developer role for work rather than a software dev. Regardless, this shit is just what i needed to know.

What the hell did I just watch?

It is just a simple case of querying 1 table, so join is trivial. Not much use when you have a complex query with many rows to return so it needs to be paged 😢

Interesting! Thanks a lot!

Thanks Erik 😀

It's unfortunate that a method that was documented 15 years ago is still a preferred method when I think that using the OFFSET ... FETCH structure is more readable/understandable (but doesn't use BATCH mode). I'll look forward to a future video from Erik Darling that demonstrates that it works now 😁

fyi you forgot to include the Paul White links in the description

Erik "Sweet Meteor of Death" Darling

Thanks 👍

02:10 Honestly, I quite liked that opening. It's rare to see a coverage of when to make a columnstore index that is neither ridiculously basic nor extremely advanced. I liked having two minutes spent on how people use columnstore.

erik darling, your Data-Darling, at Darling Data?

Cheapest investment I have ever done to improve my SQL life.

Postgres supports casting via 3 methods, and I've seen queries that combine all 3: CAST(created_at AS DATE) , created_at::date , DATE created_at

I've needed to know what a query/plan hash is before, but I've never been so unlucky to need to know what a SQL handle is. When does it come up?

Got a course on solving bad CEs mate?

This reminds me of Brent's recent video of him being unable to get Automating Tuning to work. I have no idea what he was doing wrong.

Hmmm... Is my take away, after I have enabled all of the CE Feedback features on SS2022, I should run my queries 18 times for this to kick in??? 🤣

Got a nice mad magazine influence on the new zine. Nice!

Holy sh** , it is a real zine?!?!

My query plan timed out at 6 ms 😂

Erik daaaarlin Daaaaaaarling data😂

Great stuff! Thanks! I am afraid to look at the optimizer stats at work - but Monday is coming, and I am a glutton for punishment.

If we take your video content divided by the cost (0=Free), then the value is INFINITE! Thanks for all of your video content, Erik 😊

I think it matters how you feel! You are more than just videos, you are also the top DBA as voted by Beer Gut Magazine 🍻