Team

Has anyone done any serious benchmarks comparing throughput for SQL
relational and MDB relational?

I've been doing a bunch of customizations to the foodmart database and
i am pleasantly surprised how fast MDB runs.

And i would like to find out more about benchmarks between MDB and SQL.

-Aaron

an MDB database is a really fast database.
but its not a database server.
so supporting multiple users is not so simple. bad lock process, no cache,
more bandwidth used, no log...
Imagine you have a 1million rows database, your client station MUST read
these rows to do a simple SUM, so its a huge network traffic, while a server
do it locally and just send the result other the network.

also the database format is not designed to support large databases, there
is a size limitation. (I don't remember the size, maybe 2 or 4Gb)

but in mono-user access on small databases I really love this engine!!!

I think you'll not found any benchmark which compare MDB to SQL Server.
if you talk about number of users supported, SQL Server win
database size, SQL Server win
transaction / minute, SQL Server win
processing an update or insert on a table smaller then 2Gb, MDB win (no
transactions, single user)

<aaron_kempf (AT) hotmail (DOT) com> wrote

i just mean read performance.

as in datawarehousing.

if i have a cube that is sourced from SQL and it takes 20 minutes to
run the query.. and then i can run the query against a MDB in 10
minutes..

i've just been blown away recently; im working on a slower server.. and
foodmart cube is able to run a query for like 300k records (3 seperate
partitions) in less than 45 seconds.

i just know that SQL can't do anything with 300k records in 45
seconds..

i was quite pleasantly surprised.

of course; im going to try to dig deeper, to verify that we're reading
all the records.

it just seems too fast for me to believe.

Yes, Jet is a very fast little engine. No, I don't
have any bench marks.

But I think I've read that SQL 2005 offers better
performance than 2000 for cubes.

(david)

<aaron_kempf (AT) hotmail (DOT) com> wrote