Daniel Rail wrote:

>
>
> It might have to do with how blobs are actually transferred. Each
> segment might be transferred individually, in separate packets, and
> all of the segment, not just to used portion.

No, it's not doing that... but I don't have an explanation for the slow
down either. There is one quirk that I think we still have. If you do
a SELECT ... FOR UPDATE, InterBase 6 and successors would send one row
at a time, rather than filling a buffer...

Other than that, sorry, not a clue.

Regards,


Ann