RE: Network Traffic Violations

["Paul D. Robertson" <proberts () clark net>]

On Fri, 11 Sep 1998, Woody Weaver wrote:

> At 09:06 AM 9/12/98 -0700, Matthew Gast wrote:
> > Is anybody familiar enough with cable modems to know how large an area
> > you're sharing bandwidth over?  If it's only a block or an apartment
> > complex or something relatively small, the impact of oversubscription could
> > be smaller than with DSL.
>
> One of my colleagues is getting a cable modem turned on tonight.  He was
> told he had 1100 other users in his domain!

In a domain isn't the same as on the same pipe, I've got tens of 
thousands of users in my domain, they're not all on a single network 
segment.  

Most cable systems that use hybrid fiber-coax (two way plant - not 
everyone is there yet) tend to have aggragation scaling issues at the head 
end. I don't recall the exact numbers and my notes are at work, but in 
non-specific terms (heavy apartment complexes may be a different issue) I seem
to recall it being on the order of 120-150 households.  Definitely, it's 
possible for the cable guys to scale it well, and it just remains to be seen 
if they will.  

Also, most newer systems allow the throttling of bandwidth at the user end - 
I'm not sure how this will play out as modems get into the consumer channel, 
but for the current rental units I'd guess it's a great way of limiting 
bandwidth problems in the short-term if an issue crops up.  

Given the lead time for connections, oversubscription at a head end should be 
fairly easy to see coming, but then again, I'd have _thought_ the same for 
T-3s at a local switch from two different wireline carriers until fairly 
recently *sigh*.

Paul     
-----------------------------------------------------------------------------
Paul D. Robertson      "My statements in this message are personal opinions
proberts () clark net      which may have no basis whatsoever in fact."
                                                                     PSB#9280