On Tue, 2 Jun 1998, Brian Kantor wrote: > I respectfully disagree. As more and more indexing and cross-reference > databases become available, more news systems will retrieve based on the > unique Message-ID of the article. That's future deisgn stuff, not NNTP distribution and it's been discussed for years. It suffers from the usual falilure mode you and others fall into inasmuch as Usenet is a superset of Internet connected machines. I feed some 150-200 UUCP boxes for example and there's no way they can use the scheme you just proposed. > Schemes like this are the reason that Message-ID retrieval was included > in the NNTP design. No, Message-ID was included to provide a unique identifier in order to allow servers using flooding methods to reject Message-IDs received via other paths in order to avoid on-disk duplicates. > >Apart from the CPU issues, a machine which can't cope with having > >the entire history database loaded up will basically be unusable > >for all the other readers on it, so message-ID retrieval turns into > >a denial of service attack. > > Which means that the news server needs to be fixed, not the users. > Program inefficiencies and/or hardware shortcomings are rarely a good > reason for enforcing restrictions on users. After all, in some sense, > isn't the purpose of the system to serve its users? The "news servers" in questions include just about every INN server at every small-mid sized ISP in North America along with most .edu machines. Most sites don't have unlimited budgets and Usenet is a big $$ eater which, if given the resouces it demands, will expand to fill them and then demand more. AB
