Subject | Re: [Firebird-Architect] Can we, can we, can we????... |
---|---|
Author | Martijn Tonies |
Post date | 2005-06-13T14:04:05Z |
> >>To:"better" - yes. "more important" - probably.
> >>
> >>Just raise a 'Timed Out.' exception
> >>
> >>:-)
> >>
> >>
> >
> >Very well :-D :-D
> >
> >
> >Next, should this be:
> >
> >1) a server configurable option?
> >
> >2) a database configurable option?
> >
> >3) a connection configurable option?
> >
> >4) all of the above?
> >
> >
> >Is the database owner affected? How about SYSDBA? (think of backup, for
> >example)
> >
> >
> >
> There are a few questions that must be answered before we go much
> further with this:
>
> 1. Is a timeout more important or better than a mechanism to identify
> and kill runaway requests?
Killing a runaway query either requires the application to surface this
functionality (eg: "Cancel" button) or a DBA to kill it for you.
A timeout will always work.
> 2. How exactly are we going to track execution time of a request?Hey, I dunno. I don't do engine development ;-)
But it probably has to do with the thread that's currently executing
some query? Can you determine which thread it is and record the
time it uses before getting stalled? Or will the time during stalling
count towards the limit (this would mean that a timeout set on the
client equals the total time the query can reside on the server)?
> 3. What are the implications on system performance in tracking andDunno.
> probing request execution time?
With regards,
Martijn Tonies
Database Workbench - tool for InterBase, Firebird, MySQL, Oracle & MS SQL
Server
Upscene Productions
http://www.upscene.com