home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!lhdsy1!nntpserver.chevron.com!usho92.hou281.chevron.com!hdnea
- From: hdnea@usho92.hou281.chevron.com (David Neal)
- Newsgroups: houston.general
- Subject: Re: News Feed
- Message-ID: <1992Nov17.173728.15372@nntpserver.chevron.com>
- Date: 17 Nov 92 17:37:28 GMT
- References: <X10LPQL@taronga.com> <1992Nov11.170002.18453@sun44.synercom.hounix.org> <1992Nov15.201040.6571@hounix.org>
- Sender: news@nntpserver.chevron.com (USENET News System)
- Distribution: houston
- Organization: Chevron
- Lines: 67
-
- In article <1992Nov15.201040.6571@hounix.org> jlb@hounix.org (Joel Breazeale) writes:
- >In article <1992Nov11.170002.18453@sun44.synercom.hounix.org> medley@sun44.synercom.hounix.org (Bert Medley) writes:
- >>In article <X10LPQL@taronga.com> peter@taronga.com (Peter da Silva) writes:
- >>>In article <1992Nov07.161442.16666@limbic.ssdl.com> gil@limbic.ssdl.com (Gil Kloepfer Jr.) writes:
- >>
- >>But the main problem as Gil has stated (this is a 2nd post) is that to carry a
- >>full feed requires more than 19 hours of modem time - 5+ to get the full feed,
- >>5+ to pass it on, and ~4 to provide 3-4 20% feeds to leaf sites. This leaves
- >>the 4+ hours that are left to try to connect and carry e-mail. All in all,
- >>probably more than 24 hrs.
- >>
- >>The problem is the volume of news. Many of the groups carry extraneous posts
- >>and recurring topics rather than taking the discussion offline. [...]
- >>
- >Actually... The volume is the problem only if you are using UUCP. If you are
- >using NNTP the volume problem is solved. NNTP doesn't transmit the entire
- >article -- it offers the Message-ID and if the recipient hasn't seen it then
- >it'll say 'send it' else it will refuse it.
- >
-
- Hi Joel!
-
- No, not exactly. The volume problem is still there for the first feed.
-
- You still have to GET the news the first time before you can refuse
- it the second, even with with NNTP. So, 5 hrs on the phone no matter
- how you slice it.
-
- NNTP does cut wasted bandwith in terms of duplicity. You can refuse
- an article if you already have it, unlike a compressed batch of
- news from your feed. Unfortunately, since NNTP is an interactive
- uncompressed protocol, it would not compare well to compressed batches
- via uucp for raw throughput.
-
- >I brought a site up under NNTP and found it easy to handle.
- >
- >Now... Is it possible to build a SLIP network amongst the USENET sites in
- >Houston and use NNTP? I've not used dialup SLIP, but I know nuchat is looking
- >into it.
- >
-
- You have the right idea, Joel.
-
- It's called ihave-sendme. One system blows a batch of ihave messages
- down (via uucp). The other system sends back a stack of sendme messages.
- Eventually the requested news articles (assuming they haven't expired)
- come back down the pike.
-
- [I'm sure Mr Da Silva will jump in with a correction if I've flubbed]
-
- Steve Nuchia did some _excellent_ work hacking his ihave-sendme system.
- Brilliant, really. He queued his ihaves from uunet and then looked for
- the messages to come in from one his free feeds. If he still hadn't
- gotten the articles for free in a few days, the sendmes went back to uunet.
-
-
- What I would like to see (if we really want a po-folks wan) is sort
- of a cobbled up ring topology. Who can afford a rack of modems?
- But, if you can afford two each...
-
-
-
- --
- David Neal -- hdnea@hou281.chevron.com
-
- My processes aren't dead! They're only resource impaired!
- Computationally challenged!
-