Are you issuing "tens of reads and tens of updates/ inserts" for your
ACID transaction individually from SQL client, or have you packaged
them as a single database function ?
Using the function can be much faster, as it eliminates all the
command latencies between the client and the server.
Cheer
It completely depends on a lot of factors of course, so these numbers are
meaningless.
It depends at the very least on:
* The hardware (CPU, disk type + disk connection)
* The size of the records read/written
* The presence of indices and constraints.
So, adding some other meaningless numbers to a
