Re: Q: Is really Sybase so much behind than other RDBMS?

From: Scott Gray <gray_at_voicenet.com>
Date: 1997/04/22
Message-ID: <5jia5u$58i_at_omni2.voicenet.com>#1/1


In article <335b16fb.8054667_at_news2.ibm.net>, Savas Pavlidis <pavlidis_at_ibm.net> wrote:
>We have almost settled to Sybase SQL server and replication server
>when some articles on usenet just make think again.
>
>Users compare Sybase's SQL Server 11 with other rdbms, and limitations
>of the server and old technology striked at me. The original thread I
>have seen it in comp.databases.oracle newsgroup and its title was
>Informix vs Oracle vs Sybase (etc...)
>
>According to replies on this thread.
>
>1. Sybase has only one (1) serialization level. This is very limiting,
>isn't it?

I am not sure exactly what you mean by this. With the current release of System 11, it is true that there is very little single-query parallelization...that is, a single query is always serial with respect to itself, however the server does a very good job at managing multiple simultaneous queries.

The next release of System 11 (called 11.5 or 11.G or Adaptive Server, take your pick) supports parallel query optimization, parallel index scans, index builds, and dbcc (consistency checks). So far the performance numbers look very good. 11.5 is currently in Beta testing, but I don't think that Sybase is publicizing the exact release date yet.

>2. Sybase's replication server is too cumbersome and doesnot recover
>easily from conflicts with tables that are updateable everywhere.

Truthfully, I have very little replication server experience, but from my conversations with people that have had a lot of experience with the solutions presented by companies like Oracle, Sybase has one of the most robust and recoverable solutions. The biggest problems that a lot of sites run into is that replication server is *very* complex (and very flexible), and is is really easy to dig yourself a hole if you don't know what you are doing.

As for updatable everywhere, I do know of several sites that currently succecssfully employ this feature on highly volitile data.

>3. Page level locking instead of row level locking. This is also a
>serious limitation, when everybody are able to do row-level locking.

I won't get into the religious arguments on row level locking, but I will say that I have yet to implement an application that suffered from extensive locking contention, it just takes a lot of thought be put into architecting your system prior to implementing it.

BUT, rll is in the works for SQL Server. I have read the functional specs and it looks like the implementation should settle most qualms people currently have about pll. Once again, Sybase has not announced a release date, but I am pretty sure the current thought is that it will not be released this year (but...I am far removed from engineering and marketing, so take everything I say as to release dates with a grain of salt).

>4. You can't prevent the use of logs for faster response (for example
>in a data warehouse application /EIS ).

Actually, you can prevent the use of logs...it just isn't documented externally. Also, this feature is server-wide, so it will disable logging on all databases residing on the server. Of course, without logs you can't do transactions, and you can't use triggers.

Check out the Sybase FAQ (http://reality.sgi.com/pablo/Sybase_FAQ) on how to enable this feature.

>I would like some opinions on the above, and if you have some other
>limitations in mind, that may be severe factors in choosing an rdbms.
>In my case replication is a crucial factor in choosing an rdbms, and
>users that have used replication server may contribute their opinions
>(and problems) here. Also how does it compare to Oracle's replication
>capability?

Speaking as objectively as possible (I'm a Sybase employee--but I woudn't work for the company if I didn't believe in them), I have always been very fond of Sybase's architecture. The server design is very elegant and you cannot find a more open database vendor or better connectivity solutions (e.g., you should read up on Open Server, and the posibilities it opens up).

Also, I am *very* pleased with Sybase's current direction they are taking with SQL Server (adaptive server), once again it seems better thought out and planned than most of the solutions I have seen come from the other major vendors.

As usual, though, I always recommend that you find out for yourself. You'll usually find that most database vendors will give you a trial copy of the database to help you evaluate the features.

-scott

-- 
Scott C. Gray                 gray_at_voicenet.com     "my keybard is brken"
Sybase Professional Services  scott.gray_at_sybase.com
   http://www.voicenet.com/~gray/sqsh.html
Received on Tue Apr 22 1997 - 00:00:00 CEST

Original text of this message