RE: Network Traffic Violations

[Rick Smith <rick_smith () securecomputing com>]

> At 02:41 PM 9/10/98 -0600, jrtietsort wrote:
> > Telco-ISP's will survive because the provide the same 
> > bandwidth to each user regardless of how many users
> > in your area are using the service.  I think you'll
> > find that cable modems, you are sharing bandwidth
> > with all the other houses in your area.  

Off topic, but their survival depends on their adaptability. Just about any
practical communications infrastructure is going to share resources, so
they're all vulnerable to degradation. For example, the telco/ISP
combination usually suffers from a finite modem bank size. It's too soon to
tell how much the user population must grow before typical cable modem
performance is only 10 times faster than the telco/ISP combination. Perhaps
the vendors will be able to scale up their infrastructure to keep customers
happy and throughput high. Time will tell.

At 09:58 AM 9/11/98 -0400, Ted Doty wrote:

> Also, make sure your Windows boxes aren't sharing their C:\ drive to
> Everyone, otherwise everyone in your neighborhood will be able to browse
> your Quicken files ...

I've been wondering about that, though our home has Macs as opposed to PCs,
and sharing is still through Localtalk. I didn't study the report on the
Windows problem when it first came through -- I assumed my small town
wouldn't see such a cutting edge technology so soon, and our business
customers usually invest in Real Network Connections.

I'm not familiar with how Windows sharing is implemented, but my intuition
is that our local cable modem implementation might not suffer from the
problem. The cable modem is installed on an in-home LAN of two or more
10BaseT hosts. One workstation on the LAN is allowed to use the Internet,
and its IP address is stored inside the cable modem (configured with an
HTML form). All Internet packets are NATted so the local workstation only
sees its own address on the Internet packets. The local workstation doesn't
really have a mechanism to find out its Internet address.

So, if Windows sharing uses LAN broadcast, then the LAN broadcast won't be
relayed unless the cable modem is really bone headed (not impossible, of
course). Since the local workstation can not find out its address on the
Internet, it can't fashion packets to automatically talk to other cable
modems in its "neighborhood" without some sort of broadcast.

So, does anyone remember how the reported problem worked? How does this
situation compare to it?

Rick.
smith () securecomputing com