Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Pro's & Con's on Oracle & SQL Svr?

Re: Pro's & Con's on Oracle & SQL Svr?

From: wayne <no_at_email.please.com>
Date: 29 Mar 2001 21:44:20 GMT
Message-ID: <9a0abk$snk@freepress.concentric.net>

> What you say is true. However the implication is open to interpretation.
 The
> SQL Server viewpoint is that it is more efficient (performance-wise) to
 have
> the database continually monitor itself and dynamically adjust its
 settings
> to reflect the load.

I personally do not like the point f view of " we know better than you know about your own system," but to each his own I guess. Also, Oraqcle is doing some of this automatically in the 9i upcoming version, so I guess it does not matter _that_ much.

Also, be carfeul, there is a difference between "self-monitoring and adjusting" and "hard-coded unchangeable settings." MS loves to hard-code stuff.

> With Oracle, the DBA has to "pick a point in time" and then issues
 settings
> for that point in time. While the settings for that point in time may be
> optimal, the likelihood that the database load is constant is relatively
> small.

Not necessarily. For example, there are statistics that are gathered and you are given a grand total, not a point-in-time total (ie: in-memory sorts vs disk-sorts, useful when choosing sort area size to tune for better performance).

> SQL Server, instead, continuously monitors itself and its performance, and
> dynamically adjusts its configuration. This tends to provide, overall and
> over time, a better level of optimization than doing a point in time
> optimization.

Again, "point-in-time" as you express it and imply it is all Oracle can give you (or it is all a DBA can do) is misleading. No everything is point in time.

> I do accept that it is difficult for many Oracle DBAs to believe that the
> software can do a better job by itself and without intervention, however
 the
> TPC benchmarks would appear to support this viewpoint.

Oh, let's not get into benchmarks. They are always rigged to tilt in favor of whoever is paying for the tests (Certianly MS and Oracle both included). Also, it will trigger a holy war and we should not loose objectivity here...

On the other hand, if you can make MS SQL server run faster than 1/3 of the speed of Oracle you can get a million dollars! (go to http://www.oracle.com). Received on Thu Mar 29 2001 - 15:44:20 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US