Is this claim true, or not?

[Charles K. Lee II]

On Sun, Jan 19, 2003 at 03:45:43PM -0600, Seth Dimbert wrote:
> PLEASE: let's not let this become a debate of TWC vs. Everest or such
> things... Can anyone give me the straight dope?

First off, the residential service is capped at 2000 kilobits/seconds
downstream and 384 kilobits/second upstream.  So, on a per user basis, you
definitely get more bandwidth than the typical DSL residential service.

When it comes to the "neighbors hogging the bandwidth" issue, while it can
be an issue, it usually is not.

Here's a basic overview of the way the network is setup.  There are Cisco
UBRs all over the city which service thousands of customers each.  Each UBR
has up to 4 CMTS cards and each CMTS has up to 6 upstream ports.  Each
upstream port services 1-3 nodes.  Each node services about 150-200
customers on average (crash3m may argue that I gave a higher average value
when he asked me before.  I did, but only because I was under the incorrect
assumption that there were 4 upstream ports per CMTS).  The number of
customers per node does depend on your neighborhood.  For example, at a
glance, it seems the highest concentration of customers is in southern
Johnson County.

The usual bottleneck is the available upstream bandwidth/node.  This is why
the bandwidth provided is assymetrical.

If I remember correctly, most upstream ports are configured to provide 5
mb/s.  That may sound low, but in reality it's more than adequate.  For
example, I'm in Johnson County in an area that is above average when it
comes to number of customers/node.  Over the past months, the average
upstream utilization is about 20% and the maximum recorded upstream
utilization is 40% (these values were figured with 1 day averages.. the 5
minute averages also report the same percentages).  For the curious, the
tool used to gather this info is MRTG.

However, there are a couple hot spots where the upstream utilization is
getting close to being maxed out.  There are multiple ways these situations
are dealt with (and like zscoundrel said, they aren't automatic...
unfortunately equipment doesn't install and configure itself).

One way that is used is installing another UBR for an area which usually
halves the number of customers serviced by the original UBR.  Another method
that is currently being implemented is changing the downstream and upstream
modulation rates.  The modulation rates dictate how much data is packed into
a frame.  For example raising the upstream modulation rate from 4 QAM to 16
QAM (I may be wrong about these values/acronyms) will double the available
upstream bandwidth (5 to 10mb/s, I'm pretty sure I'm correct about these
values though).  The only problem with this is that raising the modulation
can cause problems that weren't previously evident with poor cable plant.

Anyway, I'll stop rambling now.  I may have erred at points in my
explanation since I'm not a cable tech.  I just try to pick up as much
knowledge as I can at work (if you couldn't tell already, I work for TWC).

The whole point of what I said above is to illustrate that TWC does provide
a service which is generally faster than that provided by DSL ISPs.  In
addition, the network is upgraded to accomodate the number of customers we
service.

On a seperate, but related note (sorry Seth if this ignites a provider flame
war), I have never understood why people bitch about the 384k/s upstream
cap.  It's the fastest residential upstream connection I've heard of (DSL =
128k/s, Everest = 256k/s, correct me if I'm wrong!).  Yeah, it would be nice
to have a faster upstream connection, but bandwidth = money.

-- 
chuckx | Charles K. Lee II
chuckx (at) cold-sun . com
http://www.cold-sun.com