Re: What about the Oracle vs Sybase Ads?

In article <1993Feb16.142209.25725_at_mic.ucla.edu> ddruker_at_agsm.ucla.edu (Daniel Druker) writes:

>In article <des.91_at_helix.nih.gov> des_at_helix.nih.gov (David E. Scheim) writes:
>>capabilities in this arena from a vendor which up to now did not even offer
>>optimization in its basic SQL queries. -- David Scheim
 

>Hmm. Actually Oracle has contained a Syntax based optimizer for years
>now. What you mean is that Oracle didn't have a cost based optimizer.
>There have been years of debates over which is better - I feel that some
>combination of the two which also allows for smart programmers to
>override the optimizer's "Guess" is most appropriate.

With good marketing, you can put a good spin on many things. E.g., a watch that doesn't run: "It's totally accurate twice a day, while most watches are never totally accurate at all." I remember the days when NBI word processing salemen tried to convince me that their 16-line display was better than the standard 24-line displays because they were more easily readable, and had designed them such for that reason. Actually, that company, always behind in technology and creative in marketing, could not provide the memory required to handle a full screen.

>My point is, what Sybase gives you is much more like a kit than a
>production product. The whole idea of programmable server is a lot
>like saying "Here's your new system. It's really a bunch of prototypes,
>but you can make it do anything you want if you write enough C code.
>This is certainly flexible but it is proprietary, non-standard, non-
>automatic, and less efficient than the industrial strength methods
>the other vendors use. Case in Point: Anyone out there know of a single
>large application in Sybase that actually uses two Phase commit ?
>Maybe there are five or ten in the world out of how many thousands of
>installed Sybase customers. This to me is a great example of how
>pushing technology can create a lot of hype, but without useability
>no one will implement real applications.

To clarify, you're referring above to distributed capabilities. Can you tell me how many production applications are out there using any vendor's software that use distributed client-server computing? Indeed, when we did a fairly extensive review of U.S. and some foreign sites of one major unnamed vendor a few years ago, we could not find even one large client server application, let alone one involving distributed computing. Again, this is a complex technology to deploy in production. I'm afraid that whether or not the coding to effect a two-phase commit is cumbersome or simple is a small matter in the scheme of things.

/*********************************************************************/
/*                      --- David E. Scheim ---                      */
/* BITNET: none                                                      */
/* INTERNET: desl_at_helix.nih.gov          PHONE: 301 496-2194         */
/* CompuServe: 73750,3305                  FAX: 301 402-1065         */
/*                                                                   */
/* DISCLAIMER: These comments are offered to share knowledge based   */
/*   upon my personal views.  They do not represent the positions    */
/*   of my employer.                                                 */
/*********************************************************************/