Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
X002548
Not Just a Number
15586 Posts |
 Posted - 2011-03-01 : 17:19:00
|
| ...ok...So...I have a group of developers...using "pega" a silver bullet 3rd party dev tool..that stores all of the data in hieracrchal blobs in udb (morons).They want to retain all of the data in their app/database. THEN they want, enmasse to update 25+ tables in a sql server database.Now, I fight and argue and say this is insane....the "round table" seems to have no DB background at all....uses excuses like...it's only a small population of users...transaction are small, it will NEVER happen that way....So, I don't think I've ever seen such insanity, but...what can you add to say that this is dumb?Length of transaction will be too longLockingparallel threads trying to do the same updatesIdentity column problemsI went over so many bad, dark and gloomy things...and it was like looking into deers eyes on the highwayShould I just say ok?Or, fall on my sword...AGAIN....and AGAINAnyone ever hear of Pega?http://www.pega.com/Of course this VERY expensive product was bought by VP's..I would have loved o ask for a demo of the product connect to ANY RDBMSSince the money is spent, IT HAS TO WORKBrett8-)Hint: Want your questions answered fast? Follow the direction in this linkhttp://weblogs.sqlteam.com/brettk/archive/2005/05/25/5276.aspxWant to help yourself?http://msdn.microsoft.com/en-us/library/ms130214.aspxhttp://weblogs.sqlteam.com/brettk/http://brettkaiser.blogspot.com/ |
|
|
robvolk
Most Valuable Yak
15732 Posts |
Posted - 2011-03-01 : 17:33:42
|
| If the money is spent and they don't want to hear "it's not good", give up now. It's not worth the aggravation.If you want to actually fight though, have them run demos against a SQL Server and trace the statements it generates. Take that trace and translate the SQL into something that performs better. Even better, see if you can scale test your code (2x, 5x, 10x, 100x) and graph the performance. Try to get them to do a similar scale test with Pega. The best you can do is show them numbers comparing different approaches. And if they STILL don't get it, then give up. But you're at least off the hook for the shitty performance they're going to get. |
 |
|
|
X002548
Not Just a Number
15586 Posts |
Posted - 2011-03-01 : 17:51:08
|
| Thanks RobWas already going down that pathTrouble I have is that it all has to be in 1 transaction...I have no calling app to rollback a single failure to on any one tablethey could have dozen of fail points, and at only one point will they see the first one...then the fix that one, and try to insert again, which may fail further down the lineANY ONE in IT Should have at LEAST Some Database understanding...hye it's not like in the late 1990-2000 when EVERYONE (taxi drivers, Cashiers, hair dressers) tried to get in the busiess...but WHERE DO THESE PEOPLE Come from...is it the left overs because everything is being outsourced?one guy was doing SET @x = ...some XQUERY sheet..I said what happens if there's more than 1 row in the XML"Oh, they said THAT WOULD NEVER HAPPEN"Brett8-)Hint: Want your questions answered fast? Follow the direction in this linkhttp://weblogs.sqlteam.com/brettk/archive/2005/05/25/5276.aspxWant to help yourself?http://msdn.microsoft.com/en-us/library/ms130214.aspxhttp://weblogs.sqlteam.com/brettk/http://brettkaiser.blogspot.com/ |
|
|
|
robvolk
Most Valuable Yak
15732 Posts |
Posted - 2011-03-01 : 18:08:23
|
quote:
Trouble I have is that it all has to be in 1 transaction...I have no calling app to rollback a single failure to on any one table. they could have dozen of fail points, and at only one point will they see the first one...then the fix that one, and try to insert again, which may fail further down the line
Use table variables if you want to preserve stuff, they're immune to rollbacks. It's more work and kind of a dumb workaround, but if you can CATCH the error you could possibly match it up to the bad data that caused it.quote:
ANY ONE in IT Should have at LEAST Some Database understanding...hye it's not like in the late 1990-2000 when EVERYONE (taxi drivers, Cashiers, hair dressers) tried to get in the busiess...but WHERE DO THESE PEOPLE Come from...is it the left overs because everything is being outsourced?
After the last bubble, I'm convinced IT is no different than any other field...you get geniuses, and you get people who failed the burger flipping test at McDonald's, and they still need a job...quote:
one guy was doing SET @x = ...some XQUERY sheet..I said what happens if there's more than 1 row in the XML. "Oh, they said THAT WOULD NEVER HAPPEN"
Depending on how confrontational you want to be, some possible responses are:1. Who's "they"?2. How do "they" know? Did they show you the source code that GUARANTEES there won't be more than one?3. "They" are a bunch of <long stream of obscenities> idiots, and you're <an even longer stream of obscenties> idiot for believing them.4. There's a long history of experts claiming things would never happen, but did. i.e. powered flight, computers, conquering diseases, and so on. "They" were wrong. |
|
|
|
|
|
|
|
|
Update 25 tables