« first day (1521 days earlier)      last day (3347 days later) » 
00:00 - 23:0023:00 - 00:00

11:31 PM
I think someone doesn't understand the feature he is writing about
@JamesLupolt I think the confusion comes from trying to match stats_column_id to sys.columns.column_id or sys.indexes.index_column_id, no? The order in the table and the order of the index are not necessarily the same. I'd have to read your question though to see where the problem is, and I've got a thick queue tonight.
(I'll confess, I didn't read the question, just going off the discussion here.)
 
@AaronBertrand I confess I know almost nothing about sql-server statistics
... but 200M rows tables = 200M rows of statistics? Is that what he is saying?
 
@ypercube I can't quite exactly parse what he's talking about, but he's clearly just stabbed at a guess at what the feature does.
 
So sad. One of my DBA team members hasn't read Dune, the other hasn't read Hitchhiker's.
 
Has nothing to do with number of rows updated etc. Incremental statistics allow you to update the stats on a partition instead of the whole table (e.g. consider the sliding window case, where you have a "hot" partition with current data, and a bunch of cold partitions that have older, static data).
 
But it's a world of information
 
11:38 PM
So an important detail he left out is that you have to have a partitioned table and you have to know which ones have churn. And his statement "update just those rows that have changed and merge them with what's already there" makes no sense to me. You're updating the data?
 
You just edit the values using a hex editor
How's London?
 
No, you're supposed to say it's horrible and the whole sqlbits hashtag is full of lies
 
Oh, yes, sorry, nobody is having any fun here
 
These people look miserable
 
11:42 PM
@AaronBertrand I see. Searching the Net, found that even our friend Pinal has got that point (about partitions): link removed
 
I hope I'm not flagged for linking to that site ;)
 
bad cube
I kid, he has useful material sometimes, even if it's regurgitory. But when he's wrong he's really, really wrong.
 
@AaronBertrand I think I'm using the columns correctly (at least according to Books Online). If you ever have time there's an easy repro in the question. If I'm making a mistake (which is likely) I'm happy to hear where.
Have fun at SQLbits... I should be there on Saturday
My coworker Steven King is there now. And my ex-coworker Andy Hogg
 
FWIW I've filed a dozen bugs against BOL in the past week.
 
11:55 PM
Plus a few other people from Team Putin (Gazprom)
 
00:00 - 23:0023:00 - 00:00

« first day (1521 days earlier)      last day (3347 days later) »