From: leebert <*GNOSPAM*leebert@mindspring.com>
Subject: Re: MS SQL Server vs Oracle vs DB2 (&Sybase too)
Date: 2000/05/27
Message-ID: <39301862.25B1D7DE@mindspring.com>#1/1
Content-Transfer-Encoding: 7bit
References: <1214dc3c.60e30c83@usw-ex0109-068.remarq.com> <392E6F5D.2E9F1A1B@mindspring.com> <8gmjna$l8p$1@nnrp1.deja.com> <8gmrve$r94$1@nnrp1.deja.com>
X-Accept-Language: en
X-Server-Date: 27 May 2000 18:52:50 GMT
Content-Type: text/plain; charset=us-ascii
Organization: Narnia
Mime-Version: 1.0
Reply-To: *GNOSPAM*leebert@mindspring.com
Newsgroups: comp.databases.oracle.server,comp.databases.ms-sqlserver,comp.databases.ibm-db2,comp.databases.sybase


jijju@my-deja.com wrote:

> Sybase is much better compared to Oracle and
> DB2. Especially the SQL based stored procedure is good
> and comes handy.

Oracle has PL-SQL. I thought it was pretty good , as good as Sybase's T-SQL from what my
Oracle friends have told me.

DB2 ver. 7 will have SQL-PL based on a proc language standard.

> temp database is very handy to resolve many big quiries.

Let's not forget MS SQL has tempdb, w/ #tables as well. And w/ DB2, you can give the users
their own schema that serves the same function.

> Updates are powerful with joins.

I thought DB2 has these.

> Replication concept is much better in Sybase than other databases.

Really? I'm curious how that is.. MS SQL does 3 kinds of replication. I'm not impressed by
DTS, but DirectConnect ain't a breeze either.

> No other database is that
> powerful as sybase.

later self-contradicted by:

> However Db2 ver 7 come with much powerful SQL support and refined database concepts.

Yes. DB2 v. 7 is going to kick some butt.

> it is always
> not Oracles defect, It is only your poor understanding of database
> technique. You shouldn't be surprised If they ask to change either the
> machine or the DBA. Oracle is the real pain.

This is what I've heard. Arrogant.

> All are equally competent in releasing the product with bugs.

Yup. But MS doesn't document patches made to MS SQL's kernel. BIZARRE!

> two simple insert / update would result in dead lock. Hope it is fixed
> in the latest release, I didn't get an opportunity to verify.

We fixed that by going w/ 3 or 4 named caches & segments for a table... that way the
spinlock is spread out. But still a pain, yes.

But we've also seen index page contention on DB2 UDB as well. Similiar approach w/ DB2:
Keep the commit size down & add more tablespaces. Again, row-versioning (Oracle,
Interbase, Postgres) will ultimately be the best sol'n for concurrency issues, I think
ultimately all DBMS vendors will go to row versioning.

/leebert

+-----[ http://leebert.home.mindspring.com ] --------+
It is impossible to begin to learn that which one thinks one already knows. -- Epictetus
(c.55-c.135)