Thank you so much for the stream of videos as each of it is a gem of its own. Anyone finding these videos anywhere through their MSSQL journey would find themselves in a far better position to understand what works under the hood of the Sql server engine. I hope more people find this channel and I'm definitely going to make this channel a must watch for anyone who wants to get sufficiently better at understanding how to write better quality queries. If i need any expert advice, i know who to connect with 🤣
Rolls right off the tongue, like the German word for squirrel.
День назад
At least explicitly creating temp tables with a proper primary key solves so many headaches (as in the stuff you will join it on). And if you make a mistake in the joins or do a union all instead of union filling the temp table you get an error that tells you exactly what is wrong instead of multiplying all the data... But what I mostly see people do is: select into and then create index on the filled table and then only read from it once. So I get why people are told to just not do it also....
There are many upsides to that approach. Fully parallel insert and index created with full scan statistics to name my favorites.
День назад
@@ErikDarlingData Right, that nonsense... I think that it can be faster to drop an index, then insert then re-create the index, then just doing an insert is just so stupid i refused to remember it. But usually you get the same benefit without re-sorting that much.
Thank you so much for the stream of videos as each of it is a gem of its own. Anyone finding these videos anywhere through their MSSQL journey would find themselves in a far better position to understand what works under the hood of the Sql server engine. I hope more people find this channel and I'm definitely going to make this channel a must watch for anyone who wants to get sufficiently better at understanding how to write better quality queries. If i need any expert advice, i know who to connect with 🤣
Thanks for the endorsement 😃
11:33 NTEXT or XML is safest for all columns. Message brought to you by Storage Vendors.
Pure-ly coincidental.
@@ErikDarlingData wooooooah I didn’t have my helmet on for that. 🤣
It’s called 3VL. Three valued logic!
Rolls right off the tongue, like the German word for squirrel.
At least explicitly creating temp tables with a proper primary key solves so many headaches (as in the stuff you will join it on).
And if you make a mistake in the joins or do a union all instead of union filling the temp table you get an error that tells you exactly what is wrong instead of multiplying all the data...
But what I mostly see people do is: select into and then create index on the filled table and then only read from it once. So I get why people are told to just not do it also....
There are many upsides to that approach. Fully parallel insert and index created with full scan statistics to name my favorites.
@@ErikDarlingData Right, that nonsense... I think that it can be faster to drop an index, then insert then re-create the index, then just doing an insert is just so stupid i refused to remember it. But usually you get the same benefit without re-sorting that much.
The sorting can be painful. It would be nice if indexing temp tables had some shortcuts.
I totally understand picking NOT EXISTS instead of NOT IN, but I haven't seen any convincing argument for using EXISTS instead of IN.
There isn’t much of one generally, unless you are playing additional tricks with the IN clause.
…Mom…
😂