 |
 |
Archives of the TeradataForum
Message Posted: Thu, 18 Sep 2003 @ 16:49:04 GMT
| Subj: | | Re: COGNOS Impromptu Scheduler |
| |
| From: | | Frank C. Martinez IV |
Uh, no, it's worse than that.
Someone does that same query in the tool formerly known as "Queryman" and the aforementioned sunglass covered application comes back with
"Here's the first 2000 rows, do you want to see anymore, human?" (or something like that). Unfortunately the human has stepped away to fill
their belly. And gets hit by a truck crossing the street to McDonald's. Guess what's still got a read lock on it? And later than night,
the MultiLoad job waits and waits for that person to get back from lunch, answer the stupid question and release the lock. Bad.
This is why all user access should be through a view, and that view should have "locking for access" in it. Well, OK, there are
exceptions, but they are few and far between. (Note: I have seen "locking row for access, but haven't figured out what that does, as
opposed to on an update using the PI). So they can walk away all they like, and it don't make no bit o' diff.
iv
| |
