You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by Rick Strong <rs...@panix.com> on 2004/11/30 16:55:24 UTC

Derby - How Scalable?

I've been using Cloudscape for quite a while in an application that 
hardly exercises it to its limits - relatively small amounts of data, 
few time critical functions, etc.. I gather that IBM is positioning 
Cloudscape as something of an entry-level product that one might migrate 
from when demands become too great, and move to DB2. As I'm preparing to 
move from an old version of Cloudscape to Derby, I'm wondering if Derby 
might actually be able to to accommodate larger datasets and support 
faster access than its positioning might suggest. This will be a servlet 
based system with up to ~100 concurrent users, each reading/writing data 
in 5-10K byte chunks every few seconds - this needs to be done in a 
timely manner. I'm not sure yet what our size limit on accumulated data 
would be, but bigger would of course be better.

Are there any opinions as to whether such a system might need to abandon 
Derby in favor of DB2 or something else, and if so what the approximate 
limits are? All opinions welcome.

Rick Strong



Re: Derby - How Scalable?

Posted by Jan Hlavaty <hl...@code.cz>.
Rick Strong wrote:
> I'm wondering if Derby
> might actually be able to to accommodate larger datasets and support
> faster access than its positioning might suggest. This will be a servlet
> based system with up to ~100 concurrent users, each reading/writing data
> in 5-10K byte chunks every few seconds - this needs to be done in a
> timely manner. I'm not sure yet what our size limit on accumulated data
> would be, but bigger would of course be better.
> 
> Are there any opinions as to whether such a system might need to abandon
> Derby in favor of DB2 or something else, and if so what the approximate
> limits are? All opinions welcome.

There is too many unknowns (hardware, JVM used, actual  operations,
concurrency issues...).

The best way would be to build a simplified prototype of the app and see
how well it performs, preferably on similar hardware you would like to
deploy it on.

Jan