Re: Informix vs. Sybase vs. Oracle vs. (gasp) MS SQL Server

In most cases, row level locks require more resouces than page level locks. Hence having page level locks rather than row level locks will speed up your queries. There are, however, instances where row level locking is necessary (due to the nature of the transactions or the schema). This is why Sybase allows you to select different locking strategies for different objects. It is all a horses for courses issues. I have seen applications written to take advantage of page level locks that outperform row level locking. I have also seen the reverse. There has been countless arguments for both locking strategies and while they have their merits, the reality is that they are both needed.

Hope this helps,
Harnam

Jon Strayer wrote:

> Anthony Mandic wrote:
> >
> > Gary L. Burnore wrote:
 

> > > Tell the airlines that row-level locking isn't important. If the entire
> > > seating table for a flight was locked for one attendent to assign you a seat,
> > > it'd be caos.
> >
> > Thats a table lock. Any programmer who does this deserves what'll
> > happen to him/her.
>
> If table level locking is prefered over page level locking then why
> isn't row level locking preferable to page level locking?
>
> > > Row level locking is quite important in some instances. I also agree with his
> > > statement that to defend Sybase by saying that YOU don't think row level
> > > locking is important is silly.
> >
> > I thick the claims being made are specious without any hard
> > evidence to back them up. I have yet to see any. Can you or
> > anyone else provide a concrete example?
>
> It seems to me that this is the kind of question that should be answered
> in a good text book that also gives a big O kind of formula to tell when
> you need one or the other.
Received on Wed Jan 06 1999 - 00:00:00 CET