Re: bytea, jdbc, i/o ...

Kenneth Been <kennethb@telocity.com> writes:
> Here is the strange behavior:
> For database A, the difference in performance between testing
> locally and testing over the network is negligible; about 5%
> slower over the network.  About what I would expect.  But for
> database B, the difference is huge; about 35 times slower over
> the network.

That's odd.

> Here is my guess for what is going on:
> Maybe for a local connection the db engine delivers the bytea as
> is, whereas for a remote connection it converts the bytea to a
> string representation, and then Java would have to convert it
> back for the getBytes() method.

I don't believe that for a millisecond; there simply is not any
local/remote discrimination in the backend.

I'm not sure what the problem is, but I think you need to find a
different tree to bark up.

Some other thoughts that come to mind involve network saturation,
routing problems, DNS problems, SSL encryption overhead, multibyte
encoding conversion, yadda yadda...

            regards, tom lane