Re: Best aproach for multiplatform RDBMS development

One transaction a second is rather funny. I'm working with a system that is running about 250,000,000 sql transactions/day (you do the math). Granted this isn't SQL7 or Access and we're also about 70% idle (ummm, smp hardware). Oh yea, it's on ~20G of data so it's a fair amount of data (not a warehouse mind you)

The catch here is that _my_ realtime isn't necessary _your_ realtime. We don't use any stored procedures but that's just the way I like it. I hate triggers and stored procedures but that's because I'm old and still think that the database should just store and fetch data.

Our systems have a code layer between the application code and the database, this allows us to handle different databases at the code level and not with stored procedures.

bobm

btw: Teradata doesn't have _real_ stored procedures yet so be careful what database features you use if you want to be portable. (although the ESQL_C from teradata really really stinks too).

In article <19990415074046.20809.00000092_at_ngol08.news.cs.com>,   jplahman_at_cs.comnospam (Jplahman) wrote:
> In article
> <Pine.LNX.4.04.9904071012180.28952-100000_at_reflections.eng.mindspring.net>,
Todd
> Graham Lewis <tlewis_at_mindspring.net> writes:
>
> >"real time performance"? Really?
>
> Please define real time. The reason I ask is we receive data (1200 integer
> variables) from a plant control system at 55msec (or 18 samples /sec). We are
> just now looking at converting our legacy application to a database like
Oracle
> because we were always told that a database engine cannot keep up with this
> high rate of data. The fastest it can handle data is one transaction / sec.
>
> JPL
>
> jplahman_at_cs.com
>

-----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own Received on Mon Apr 19 1999 - 15:41:36 CDT