ietf-nntp Backfill: not again.

[Ade Lovett]

Alan Barrett writes:
>
>I am less sure about inventing something new to help servers (and site
>administrators) avoid backfill; perhaps the most we could do in this
>group would be to recommend that some other group designs a suitable
>mechanism.

Exactly.  We seem to have a good consensus about backfill (ie, its
not permitted) so thsi should mark the end of that particular issue
(at least as far as this working group is concerned).


>> If the volume of news gets too great for a single feed to a slave
>> server, "tough luck"
>
>Yes, that's one possible decision that this group could make.  If we do
>so, it should be clearly documented.  (And I suspect that some sites
>would ignore it, but that's their problem.)

Sadly, in the wonderful commercial world, saying things like
'tough luck' isn't particularly useful.  Looking at the feed statistics
from Demon (http://www.paranoia.org/news/), the daily size of the
newsfeed trebled from January '96 thru December '96 -- whether the
growth continues in such a manner is anyones guess, but it does mean
that, sooner or later, the "tough luck" point will be reached by more
and more sites, who find themselves caught between a rock and a hard
place, when dealing with replicated newsbases.

If there's sufficient interest, I'll quite happily set up a mailing
list here for more in-depth discussion on master/slave replication,
thrashing out the issues associated with not only the replication
part, but also keeping the replicated servers compliant with RFC977bis
(no-backfill et al.)

-aDe

-- 
Ade Lovett, Demon Internet Ltd.