Re: Performance and reliability of c-tree vs. Sybase or Oracle

Josef Holzer wrote:
>

 [ ... ]
>
> Could anybody tell me, if this kind of task can be done by a Sybase or
> Oracle database with the same performance as by a c-tree database or
> even better?
> Do I have to take care of some special design considerations?
>
> Thanks for your help
>
> Josef
> email: holzer_at_teletrader.com

As other posters have pointed out, where the data collector is faster than the data storer, double buffering (usually to a flat file) is the solution, proved the storer can eventually catch up. Secondly, the matter of transaction logging imposes performance losses.

But transaction logging can be stopped on several major RDBMSes, and you can still have a backup/recovery strategy, and at-least-as-good confidence in the system as CTREE.

To answer your question about special design considerations:

What's often missed by novice users of RDBMSes, especially complex ones like Oracle, etc., is the importance of tuning. There are so many different tuning options once you look, and it is somewhat of a black art to understand. Without attention to tuning, programmers get really bad performance, and hate SQL for ever. With attention to tuning, they still dislike SQL but admit the speed is workable.

I would point out that the difference between a badly or non-tuned system and a tuned one are quite considerable, but tuning requires study (of manuals) if you're new to the area. Tuning also takes time and effort to get the tweaks and pokes right.

Tuning applies everywhere - server, disk, O/S, clients, programmatic code, SQL, client-server comms, everywhere.

• Nigel

-- 
------------------------------------------------------------------------------
Nigel McFarlane.  TUSC Computer Systems 		       nrm_at_tusc.com.au
		  666 Doncaster Rd, Doncaster 3108	   BH: +61.3.9840.4476
		  Melbourne, Victoria, Australia.	  GMT +10 or +11 hours

   When was the last time you were that frustrated _away_ from a
computer?