July 10, 2009 at 1:29 pm
I agree with what has been said so far. I am fearful of things that use words like always and never. I'd love to put my .mdf and .ldf files on different drives so long as those drives are on different controllers. Putting tempdb on its own drive may boost performance for some of us. Would you also say to put it on an SSD? Why not?
Cursors are evil. No they are not. Sorry, Jeff. Using other techniques yields better performance you say? OK. Case demonstrated. Thanks Jeff.
Every table has to have an autonumber column which has to be the primary key and is your clustered index. I not only drank that Kool-Aid, I served it to others. :w00t:
I had written stuff to shrink log files. I got very good at doing it. Then I learned better ways.
There are SQL gurus but not SQL gods. If you got Jeff, Joe, Andy, Phil, and Steve in a a single room the angels would not commence singing and no sparkly glowing would start. I like all these guys and have learned a bunch from them.
ATBCharles Kincaid
July 10, 2009 at 1:36 pm
Would be easy enough to create a sparkly glow... there's got to be enough tritium in the world for that! (Sorry, couldn't resist.)
- Gus "GSquared", RSVP, OODA, MAP, NMVP, FAQ, SAT, SQL, DNA, RNA, UOI, IOU, AM, PM, AD, BC, BCE, USA, UN, CF, ROFL, LOL, ETC
Property of The Thread
"Nobody knows the age of the human race, but everyone agrees it's old enough to know better." - Anon
July 10, 2009 at 1:52 pm
The title of this thread reminds me of this book:
Real Americans Admit: "The Worst Thing I've Ever Done"
http://www.amazon.com/Real-Americans-Admit-Worst-Thing/dp/1561631574#
Maybe we need a new thread: "The Worst Advice I've Ever Given"
July 10, 2009 at 2:37 pm
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".
😉
July 10, 2009 at 2:46 pm
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".
That one only applies to the Compact Edition, IMHO. No, I mean for doing it. :crazy:
ATBCharles Kincaid
July 10, 2009 at 2:52 pm
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".😉
Ha! I'll have to second that one. Not just in forums, but this one runs rampant in development shops.
July 10, 2009 at 2:57 pm
John Rowan (7/10/2009)
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".😉
Ha! I'll have to second that one. Not just in forums, but this one runs rampant in development shops.
I've also heard it in Classes at the University level. I think part of it depends on what you actually considr normalization/denormalization.
July 10, 2009 at 2:57 pm
John Rowan (7/10/2009)
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".😉
Ha! I'll have to second that one. Not just in forums, but this one runs rampant in development shops.
When people say they “Denormalized for Performance”, that’s code for “I have no idea how to normalize a database, so when someone complains that the database isn’t normalized, I say it was Denormalized for Performance.”
July 10, 2009 at 3:06 pm
Michael Valentine Jones (7/10/2009)
John Rowan (7/10/2009)
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".😉
Ha! I'll have to second that one. Not just in forums, but this one runs rampant in development shops.
When people say they “Denormalized for Performance”, that’s code for “I have no idea how to normalize a database, so when someone complains that the database isn’t normalized, I say it was Denormalized for Performance.”
Too true... all you have to do is ask to see the "fully normalized" design that it was derived from -- there never is one.
July 10, 2009 at 3:14 pm
cs_troyk (7/10/2009)
Michael Valentine Jones (7/10/2009)
John Rowan (7/10/2009)
cs_troyk (7/10/2009)
"Denormalize for performance" and its sibling "You only need to normalize to 3NF".😉
Ha! I'll have to second that one. Not just in forums, but this one runs rampant in development shops.
When people say they “Denormalized for Performance”, that’s code for “I have no idea how to normalize a database, so when someone complains that the database isn’t normalized, I say it was Denormalized for Performance.”
Too true... all you have to do is ask to see the "fully normalized" design that it was derived from -- there never is one.
Exactly, so asking the question of how much performance improvement they got from the 'denormalizations' usually brings a dumbfounded look. These types of 'denormalizations' are, as you said, not derived from a fully normalized model; rather, they are guessed at and justified by saying it was done for performance. I don't even like the term denormalized. It is either normalized or not! I think I am going to start using the phrase 'Spreadsheetized' in place of 'denormalized'.
We 'Spreadsheetized' it for performance!
July 10, 2009 at 4:06 pm
All too often with security issues on Production environments or time constrained developments, they will be fixed by adding the application user account to Local admin on the SQL Server or all servers in the solution.
This drives me bonkers! It's usually laziness or lack of knowledge to fix the issue. Even when the answer is the quick win and 'lets revisit later' it never gets done.
Steve.
July 10, 2009 at 4:11 pm
John Rowan (7/10/2009)
[...] Spreadsheetized [...]
Have you copyrighted the term yet?
Are you going to charge royalties for its use?
Just so 😎
And encapsulates the gist of most of the bad advice I've seen handed out, and handed out myself :blush:
Peter Edmunds ex-Geek
July 10, 2009 at 4:16 pm
Steve Newton (7/10/2009)
All too often with security issues on Production environments or time constrained developments, they will be fixed by adding the application user account to Local admin on the SQL Server or all servers in the solution.This drives me bonkers! It's usually laziness or lack of knowledge to fix the issue. Even when the answer is the quick win and 'lets revisit later' it never gets done.
Steve.
There's never enough resources to do it right the first time, there's always enough resources to fix it once it gets broken in production.
Peter Edmunds ex-Geek
July 10, 2009 at 4:18 pm
Here is a great story about a "Spreadsheetized" design where all the companies data was stored in a single table.
Death by Delete
http://thedailywtf.com/Articles/Death-by-Delete.aspx
"...
One day, a developer was optimizing the database and removing records that MegaPetCo no longer needed. All it took was a single, poorly formed delete query to wipe out each and every row in the database table.
To say the developer panicked is akin to describing a quadruple amputation as a mere flesh wound. MegaPetCo's sales immediately ground to a halt. Along with everything else. Payroll, logistics, reporting, purchasing, you name it. Its Web site didn't work -- but that was the least of the company's worries. The status of its backups was bleaker still: None. Zip. Zilch. Nada.
...
...
Unfortunately, the single delete query proved to be far more than MegaPetCo could bear. Within a few months, the company filed for bankruptcy and was forced to close every one of its stores -- laying off several hundred people along the way."
July 10, 2009 at 4:25 pm
wldhrs (7/10/2009)
Steve Newton (7/10/2009)
All too often with security issues on Production environments or time constrained developments, they will be fixed by adding the application user account to Local admin on the SQL Server or all servers in the solution.This drives me bonkers! It's usually laziness or lack of knowledge to fix the issue. Even when the answer is the quick win and 'lets revisit later' it never gets done.
Steve.
There's never enough resources to do it right the first time, there's always enough resources to fix it once it gets broken in production.
You aren't one of our dev guys or vendors are you? 🙂 Being the guy that has to fix it I guess I am always going to feel like this.
Viewing 15 posts - 16 through 30 (of 52 total)
You must be logged in to reply to this topic. Login to reply