Garrett D'Amore writes:
> Peter Memishian wrote:
> > Documentation is useless as a programming interface, which is what e.g.,
> > Fishworks needs this for.  The application can account for its own
> > overheads and provide whatever clues are necessary to help the driver work
> > together to find the right value, but leaving it to the documentation just
> > doesn't cut it.
> >
> > --
> > meem
> >   
> How do you programmatically handle this?  8150 is almost always 
> *wrong*.  The only thing its good for is a benchmark special, or an 
> isolated network with just these kinds of devices on it.  Everyone 
> everywhere else will use 9000 as that is the de-facto standard.

Indeed.  I think the intended usage is with networks that are even
more of a "walled garden" than your garden-variety walled garden.  ;-}

The question is: for those who don't care about normal
interoperability and *do* care about wringing out the last bit of
performance from these drivers by customized network-wide tuning (one
hopes that _all_ the other avenues of performance improvement have
already been explored, so we're not just talking about mere hackery),
how do we make this information available?

I still agree with you, Garrett, that this is fundamentally a
documentation issue, because the administrator of most such systems is
almost certainly going to need to deal with multiple devices with
different capabilities and configuration needs.  That means "planning"
-- a task that, in networks at least, the CLI isn't well suited to
solving.

If we really want to automate it, then I think the right answer is to
make the performance details available to a network management tool
that will help the user meet all of the constraints of his network.  I
still think that implies that exposing this on the command line will
be at best misleading, and at worst actively harmful -- it'll send
confused users down some dark alleys.

It's a bit like trying to "tune" a disk for the optimum transfer size
based only on the disk's parameters, and without thinking about
whether the OS supports that size or what sort of workload the
application will offer.  Great ... it works best with 9371 bytes per
transfer.  Now what?

-- 
James Carlson, Solaris Networking              <[email protected]>
Sun Microsystems / 35 Network Drive        71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to