|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
|
Re: usenet (was: DCAnet (was: Re: [PLUG] Comcast Crackdown)
|
On Thu, 07 Feb 2002, jaw+plug@tcp4me.com wrote:
>
> | The biggie is outgoing bandwidth since every article just comes in once, but
> | goes out many times -- to the peers and to whichever users want to download it.
>
> not really.
> you mentioned that your in+out were about the same.
>
> while you *could* send the article out to each and every peer, if you have
> reasonable peers, you won't (because they have other peers sending them
> articles, also). on an typical news server, that has similar peers, the in
> and out will be roughly the same.
In and out to the peers may be roughly the same. I really don't have the
downloaders that some place like Giganews or Supernews has.
Jeremy Nixon and some of the other very Big New Sites have said that users
(which none of the local ISPs have boatloads of on broadband) take up multiples
of bandwidth compared to feeds, if I'm remembering the discussions correctly.
Giganews just signed a deal with Comcast -- there was enough serious
complaining by Comcast users about not having Usenet that Comcast is offering
the lowest end Giganews package free and Jonah plans to make it very easy for
Comcast users to upgrade their systems.
Cox Cable went with Supernews.
My users probably post more than your average users did at op.net (I believe at
one point I was the only person posting more than one or two posts a week). I
know I was the only person who noticed when Cynet's main news server
(news1.alterdial.uu.net) was UDP'ed. UU.net security people said I was the
*only* person who called them.
Most people use Usenet to download binaries. I know my users post more than
op.net's users did. I could be wrong, but I wouldn't be surprised if they
didn't post more to text groups than Netaxs dialup customers (they certainly
post more to the groups I read but that's how I found them).
>
> in fact, ignoring end-user traffic [1], the aggregate across all Usenet sites,
> the in + out will be _exactly_ equal.
I wouldn't ignore end-user traffic now that those people have T-1 equivalents
for downloading and utterly no reason not to, and news reader programs that do
things automagically.
It's that end-user traffic that I'm referring to. I/O with peers is equal, but
if one has a lot of broadband bunnie suckers, then outgoing is going to be
quite a lot more than incoming, not that the NSPs don't charge by the megs
downloaded now unless they're speedcapping.
The only spikes I get is when someone decided to open control.cancel and get
all of them so he can grep for rogue cancels (yeah, I allow this).
>
> [1] and the tiny trickle being pushed in various multi-cast
> experiments/projects
Now that stuff is really is insignificant.
--
Rebecca Ore
http://www.ogoense.net
______________________________________________________________________
Philadelphia Linux Users Group - http://www.phillylinux.org
Announcements-http://lists.phillylinux.org/mail/listinfo/plug-announce
General Discussion - http://lists.phillylinux.org/mail/listinfo/plug
|
|