> I did not manage to speed up IBO... but I did manage to to
> slow down FIBPlus:

My mistake... again. Did you notice the record count was wrong
for the fibplus transfer - it was still set on the old table.

The actual times were:

= = = xfer_fibplus = = =
App Init Time: 0ms
Object Prepare Time: 78ms
Records Transfered: 106
Records Transfer Time: 8282ms
Transaction Commit Time: 1156ms
Connection Close Time: 15ms
Application Run Time: 9531ms

compared to IBO at:

= = = xfer_ibo = = =
App Init Time: 16ms
Object Prepare Time: 78ms
Records Transfered: 106
Records Transfer Time: 21328ms
Transaction Commit Time: 594ms
Connection Close Time: 0ms
Application Run Time: 22016ms


Jason you should be able to see this for yourself with the
two command line apps I posted to the list yesterday. Just
set them to point to a table containing large binary blobs
and watch the CPU go crazy during the transfer. It seems
that IBO is taking more than twice as long (as FIBPlus) to
transfer these largish binary blobs (69Mb over the 106 rows).

It could be as simple as buffer sizes - time to play around
a bit and see what effects we can get.


Hans that subtype 512 is still a concern.

--
Geoff Worboys
Telesis Computing