[mythtv-users] Way OT: High-speed Data Service Delivery

[Jay Ashworth]

----- Original Message -----
> From: "Jason Long" <jlong at jasonlong.com>

> Well, you eliminated a chunk of my original message which makes it look
> like I said something that I didn't, and prefaced your email based on
> that. In context, I said fiber from LEC to prem via OCx, terminating to
> a MUX, where services could be provisioned and handed off to customers..

You most certainly didn't say anything like that. 

Here's every word you said:

=============
That's not correct. A DS3 is delivered via copper coax. The fiber is
from the LEC (local exchange carrier) to a MUX which services the
location, from where the carrier can provision off a variety of services
to 1 or more customers, such as a DS3. Depending on the planned
forecast, this fiber link could be an OC3 or something larger like OC48
or OC192. Sure, once your bits hit that fiber they travel close to the
speed of light, but your DS3 signal still only allows you to transmit
bits onto the wire at a line rate 44.54Mbps (after signaling overhead),
so it's a moot that fiber is even involved.
=============

"from the LEC to a MUX" does not at all imply that you believed the
MUX was on the customer prem, nor does anything else in that paragraph.

I will admit, though, on re-reading it, that your phrasing's a lot more
equivocal than I original thought it was. :-)

What I cut, BTW, was a side trip that made it sound like velocity factor 
had something to do with bitrate, somehow.  Since the speed of light has
nothing to do with this, I didn't bother to quote it.  Trimming quotes
is a 30 year tradition on technical mailing lists.  ;-)

> A DS3 is handed off via copper coax (not fiber). I never implied
> pulling copper from LEC to customer.. and if you have 10baseT, it's
> not a DS3 anymore.

It could be metro area ethernet over a clear channel DS3, sure.  I wouldn't
bet that more than 50% of the things that use a "DS3" care about the framing.

> My point though is that fiber doesn't magically make a DS3 capable of
> serializing bits onto a wire any faster.. 

And I never said it did, as you seem to be implying.

I was replying solely to the point you appeared to me to be making, 
about physical delivery.

>                                            Further to my point is that
> since fiber propagates bits at ~speed of light, the delay introduced is
> negligible when calculating an answer to the question "How long to
> transfer CentOS over a DS3".

Sure; how you deliver your DS3 is mostly immaterial to the gain-bandwidth
product of the link.  But I was talking about layer 1, not layer 5.

> To answer the question, based on serialization delay alone, for a DS3
> w/out any L2-L7 protocol overhead, it would take ~17 minutes to transfer
> CentOS 5.42G ISOs. After adding in protocol overhead for IP and TCP it
> would take another 30 seconds. Then there is TCP slow start and sliding
> windows to account for, packet loss, re-transmissions, and L2 framing
> overhead, application layer overhead, queuing delays and current circuit
> utilization. The fact that fiber is even involved is *largely*
> insignificant for calculating a sufficiently accurate answer to the
> question.

Which is why I marked my initial reply OT: I wasn't even talking about
*that*, much less Myth.  :-)

> Back to mythtv discuss?

Oh, sure.  

Cheers,
-- jra
-- 
Jay R. Ashworth                  Baylink                       jra at baylink.com
Designer                     The Things I Think                       RFC 2100
Ashworth & Associates     http://baylink.pitas.com         2000 Land Rover DII
St Petersburg FL USA      http://photo.imageinc.us             +1 727 647 1274