If network providers implement a form of QOS that includes a scavenger
class in their network it will be very temping to direct competitive 3rd
party products into it. Any traffic placed into a scavenger class will
receive less then best effort delivery. A major problem being in either
case how can one prove that poor call quality is the result of best
effort or less then best effort? From a competitive point of view it
makes most sense for a provider to send competitive traffic to scavenger
while publicly claiming poor call quality is the result of best effort.
The definition of "best effort" becomes critical and I guarantee the
lawyers can turn a paragraph into 50 pages on what best effort is or is
not. Let's not forget the customers. They are paying for "access to the
Internet" which is poorly defined. What is included in "access"? We need
a better definition so that we can understand what "tiered" Internet
access would really mean. Does "access" mean best effort service for all
Interconnected networks? Will "tiered access" mean that content provider
paying a network access priority fee will receive a guaranteed bandwidth
into my home? There is only so much real estate on my residential
services so what are they really selling? Can they both sell priority to
my home while at the same time oversubscribing my link?
I am very interested in the technical details because it feeds back into
what is and is not damaging to free market principles. There are lot of
tools in the QOS bucket they could offer to content providers. For
instance they could offer discard eligibility tweaking.. At given level
of congestion on network port the provider can control which traffic is
more likely to be dropped. This discard threshold is an easy way to
provide better then best effort to traffic without committing to a
particular traffic level. To further define this example:
Congestion Level 1: Any traffic without QOS marking can be considered
for random discard (DSCP=0). This could included Vonage and everyone
else that does pay them beyond mutual peering.
Congestion Level 2: Consider all traffic from level 1 plus Google.com
(assuming they paid) and all others marked for level 2
Congestion Level 3: Consider all traffic for random drop including any
native services on the providers network.
In this simple example the carriers sell QOS without guaranteed latency,
jitter or bandwidth. They only have ensure that your traffic will get
dropped in the event of congestion only after other non-prioritized
traffic get dropped first. This is only one option available. If anyone
has any direct exposure to what network providers are proposing share it
with the list.
- Dustin -
Jim Henry wrote:
I think the only fair way to treat VOIP is for a provider to prioritize
their own VOIP packets, not lower the priority of VOIP packets from other
providers, or worse, block ports that competitors use for the service. That
way if I own a network I can fairly insure QOS for my VOIP customers and
give all competitors "best effort" service just like any other data
traversing the network.
Jim
-----Original Message-----
From: Hammond, Robin-David%KB3IEN [mailto:[EMAIL PROTECTED]
Sent: Sunday, March 19, 2006 3:20 AM
To: [EMAIL PROTECTED]
Cc: Dana Spiegel; nycwireless@lists.nycwireless.net; Jim Henry
Subject: Re: [nycwireless] New Yorker Article [was:
Multichannel News -AnalystsQuestionBellInvestments]
I realy dont see the need for an ISP to promote one set of
voip over another as a matter of course. How does it serve
any of the stake holders?
Granted there may be times of crisis when demand is very
high, and there is not enough pipe to go around. Any fool can
see that priority should be given to emergency calls exchange
'999' and 'x11' in these cases. The unwillingness of verizon
to allow anyone access to the 911 system results in me having
to dial around it most of the time, i often call my local
precinct on its 718.xxx.xxxx number...
I would say that non-emergency voip links should be given
round-robin priority, such that a user who picks up every
minute and hits redial will soon get through regardless of
who the voip carrier is, remain network neutral. Granted
there may be a higher bandwidth cost of routing some other
companies voip packets rather than using your own compressed
data streams, some disparity may be in the interests of all.
Ultimately some segment of the market is likely to demand
neutrality of providers in the end. But it would be nice to
be a consultant in a position to point a client company to an
ISP and say, these guys are commited to as level a playing
field as servs everyone's interests. EULAs that prohibited
use of wireless technology prevented me from recomending
verizon or cablevision for example.
What I am truly against is the practice of failing to promote
a 'rival' voip packets to provide QOS when QOS will not
threaten network capacity. Or worse yet, expressly delaying
or mangling the rival voip packets. This subtle sabotage is
unlikely to do anyone any good. The average consumer is
likely to be driven away from voip, because the issues
involved are too complicated to deal with. With less VOIP
demand, there will not be the increase in bandwidth demand
that might be spured by widespread adoption of voice and
subsequently video over IP.
In short network non-neurtrality (network hostility) is an
ill-wind that blows no one any good.
By publicly considering making non-neutrality Standard
Operating Procedure some large polygopolies are tempting
legislation that restricts the way in which all ISPs are able
to do buisness. Outside restrictions on the way one does
buisness never seem to help. If nothing else: Laissez Faire,
laissez aller, laissez passer. By abusing or considering the
abuse of a freedom that they have always had large telcos
jeopardise that very freedom. Surely this cannot be good for
anyone's bottom line?
--
NYCwireless - http://www.nycwireless.net/
Un/Subscribe: http://lists.nycwireless.net/mailman/listinfo/nycwireless/
Archives: http://lists.nycwireless.net/pipermail/nycwireless/
