Icon View Thread

The following is the text of the current message along with any replies.
Messages 1 to 3 of 3 total
Thread Benchmark program to time adding 20 million rows to table
Tue, Oct 9 2007 9:54 PMPermanent Link

Dave Harrison
All databases will work fine with 100,000 rows. But what happes when you
try and add several million rows to a table? Will the database breeze
through it and load the data lickity split? Or will it shake, rattle,
and crawl?

This benchmark program will compare the time it takes to add 1 million
rows to an empty table. It compares Paradox, DBISAM, ElevateDb, NexusDb,
and Advantage. If you're brave, I mean really brave, you can try and add
20 million rows. But make sure you have plenty of disk space and coffee
on hand. Smile

If you can improve on these times by at least 25%, then please post the
results on this thread. The object is to try and tweak the database so
it can load 10-20 million rows as fast as possible.

Are you up to the challenge?

Dave



Attachments: BenchmarkBuildTable.zip
Mon, Dec 10 2007 4:12 PMPermanent Link

David
Thanks Dave, Im insterested in any other elevatedb and nexus comparisons you
have done.
Thu, Dec 13 2007 12:18 PMPermanent Link

Dave Harrison
David wrote:
> Thanks Dave, Im insterested in any other elevatedb and nexus comparisons you
> have done.
>

I'd be interested in seeing if anyone can change the buffer settings for
the EDB table in the benchmark to get better performance out of it. So
far I haven't noticed any difference even when setting buffers to 32M or
anything in between. I don't think buffers make a difference to the
speed of inserting data into a table. The speed of DBISAM was slightly
faster than EDB. I was hoping EDB would be several times faster than
DBISAM, but that wasn't the case.

Dave
Image