I'm looking for a database system that can be used
in a clustered enviroment (just a whole bunch of Linux machines),
that has a very high performance and preferably could run as a
in-memory database system (due to much higher performance possiblities).
The main use of the cluster will be dataprocessing, so every node should
be aware of all the data and changes in the data, by any
means of replication. The system eventually should come close
to an distributed memory system. All suggestions are welcome.

Bram

"Bram Wiekens" <wiekens (AT) astron (DOT) nl> wrote

Jim

Preferably a somewhat cheaper and less complex solution,
since the cluster will consist of a few hunderd systems,
which only process a relatively small amount off data.
Process time from the outside isn't very important access
time on updates and updating in general have to be quick.

Bram

Jim Kennedy wrote:

"Bram Wiekens" <wiekens (AT) astron (DOT) nl> wrote

You are asking for a complex thing. One mistake you have made is formulated
the solution as a problem instead of stating the problem. Your solution is
probably overly complex. Heck, in a well written app I've gotten 1,500 of
users in OLTP using a 4 way windows box and Oracle.(10 years ago). Lots of
disk, but only 4 Gig ram. Nothing fancy, just tuned the thing (app and db,
the app being by far the more important part) I am sure hardware has made
some improvements in 10 years.
Jim