Someone was kind enough to forward me an outline of the peering policy
they received from Sprint. It seems to be the toughest one I've seen -
=OC48 in 14 US cities, >=OC12 to Europe. To find out the traffic
minimums and ratios it looks like an NDA is required...
Ralph Doncaster
principal, IStop.com
Ralph,
Two points, here.
One is, Sprint won't peer with you. I'm not even sure who you work for, but
rest assured, they will not peer with you. Time spent on this might be
better utilized reading some of Bill Norton's excellent intro to peering
papers, or, if you work for a company that is truly serious about peering,
you could hire an experienced peering engineer, to assist you with this.
There are several that are currently available.
The second point is, whomever you spoke to has violated a non-disclosure
agreement, one that is normally taken seriously. I would tread carefully in
this area, as it may get whomever you spoke with in a significant amount of
trouble.
Finally, I'm not sure why anyone would want to actually waste the time on
this - there are numerous large tier 2 networks that are starting to get
peering initiatives going. If you are large enough, I'm sure they would love
to peer with you. Remember - peering that first 50% of your traffic is not
that hard, if you have the resources, contacts and knowledge. It's that last
bit that hurts.
- Daniel Golding
I know Sprint won't peer with anyone small. I said in my inital post I
don't stand a chance - but who knows, at the rate OC48 prices are dropping
maybe next year I will meet the requirements. Attention K-Mart shoppers,
WorldCom OC48's on sale in isle 5. 
The second point is, whomever you spoke to has violated a non-disclosure
agreement, one that is normally taken seriously. I would tread carefully in
this area, as it may get whomever you spoke with in a significant amount of
trouble.
I didn't post this until now since I was waiting for a couple opinions to
verify that it was in fact genuine, and as well a public filing that
anyone could get if they know where to dig at the FCC.
http://ns.istop.com/~ralph/2000-04-13-sprint.pdf
-Ralph
Date: Thu, 27 Jun 2002 14:57:59 -0400 (EDT)
From: Ralph Doncaster
I know Sprint won't peer with anyone small. I said in my
inital post I don't stand a chance - but who knows, at the
rate OC48 prices are dropping maybe next year I will meet the
requirements. Attention K-Mart shoppers, WorldCom OC48's on
sale in isle 5.
Or, like Daniel mentioned, you could focus on those where you
stand a chance of peering. Note that traffic requirements mean
the more you peer with little guys, the harder it will be to peer
with their upstreams.
Pick a strategy and run with it. Being a broadband provider in
SE Canada, I suggest sniffing out public peering in NYC and CHI
for a start. IIRC, Hotmail and Y! are at AADS and will peer with
most anyone -- there's a good chunk of traffic. Now Akamaize
your network if you haven't already.
It's fun to talk about infrastructure, but there comes a time
where you draw the line, deciding what is real and what isn't...
Eddy
I know Sprint won't peer with anyone small. I said in my inital post I
don't stand a chance - but who knows, at the rate OC48 prices are dropping
maybe next year I will meet the requirements. Attention K-Mart shoppers,
WorldCom OC48's on sale in isle 5.
When OC48s are cheap, the peering requirements will become OC192. It has
nothing to do with the ability to support the traffic exchanged, but
everything to do with excluding you from peering with them.
A lot of networks are now running peering requirements so hard that almost
none of their existing peers would qualify (including some well known
tier 1's).
> The second point is, whomever you spoke to has violated a non-disclosure
> agreement, one that is normally taken seriously. I would tread carefully in
> this area, as it may get whomever you spoke with in a significant amount of
> trouble.I didn't post this until now since I was waiting for a couple opinions to
verify that it was in fact genuine, and as well a public filing that
anyone could get if they know where to dig at the FCC.
Note the date, February 8, 2000. Over 2 years later is an eternity in
peering requirement land.
>>
>> I know Sprint won't peer with anyone small. I said in my inital post I
>> don't stand a chance - but who knows, at the rate OC48 prices are dropping
>> maybe next year I will meet the requirements. Attention K-Mart shoppers,
>> WorldCom OC48's on sale in isle 5.
>
>When OC48s are cheap, the peering requirements will become OC192. It has
>nothing to do with the ability to support the traffic exchanged, but
>everything to do with excluding you from peering with them.
>
>A lot of networks are now running peering requirements so hard that almost
>none of their existing peers would qualify (including some well known
>tier 1's).
This is not news. Remember when AGIS had peering requirements that AGIS could not meet?
The obstruction to the AGIS ASN was removed by the bankruptcy courts.
<snip>
This is not news. Remember when AGIS had peering requirements that AGIS could not meet?
<snip>
Date: Thu, 27 Jun 2002 16:30:06 -0400
From: Patrick W. Gilmore
This is not news. Remember when AGIS had peering
requirements that AGIS could not meet?
Which seems to have translated into "remember AGIS?"
Eddy
Pick a strategy and run with it. Being a broadband provider in
SE Canada, I suggest sniffing out public peering in NYC and CHI
for a start. IIRC, Hotmail and Y! are at AADS and will peer with
most anyone -- there's a good chunk of traffic. Now Akamaize
your network if you haven't already.
I'm already working on NYC. For Chicago an OC3 connection to AADS would
cost more than the long-haul back to Toronto. I also prefer ethernet
peering instead of ATM. Equinix looks like the best option in Chicago,
but most long-haul carriers are in 600 Federal and not 350 Cermak.
As for Akamai, I've found very little of my traffic (<5%) is
Akamai. Akamai is at NYIIX anyway, so once I get into 25 Broadway I'd
peer with them.
-Ralph
>> Pick a strategy and run with it. Being a broadband provider in
>> SE Canada, I suggest sniffing out public peering in NYC and CHI
>> for a start. IIRC, Hotmail and Y! are at AADS and will peer with
>> most anyone -- there's a good chunk of traffic. Now Akamaize
>> your network if you haven't already.
>
>I'm already working on NYC. For Chicago an OC3 connection to AADS would
>cost more than the long-haul back to Toronto. I also prefer ethernet
>peering instead of ATM. Equinix looks like the best option in Chicago,
>but most long-haul carriers are in 600 Federal and not 350 Cermak.
>
>As for Akamai, I've found very little of my traffic (<5%) is
>Akamai. Akamai is at NYIIX anyway, so once I get into 25 Broadway I'd
>peer with them.
Mind if I ask how you got the traffic stats for Akamai?
Date: Thu, 27 Jun 2002 20:00:11 -0400 (EDT)
From: Ralph Doncaster
For Chicago an OC3 connection to AADS would cost more than
the long-haul back to Toronto.
Given as an example. However the cost/benefit plays out...
I also prefer ethernet peering instead of ATM.
Agreed.
Equinix looks like the best option in Chicago, but most
long-haul carriers are in 600 Federal and not 350 Cermak.
For now, anyway.
As for Akamai, I've found very little of my traffic (<5%) is
Akamai.
Interesting. If you don't mind me asking, what is your total
traffic level, and how did you determine 5%?
Eddy
>As for Akamai, I've found very little of my traffic (<5%) is
>Akamai. Akamai is at NYIIX anyway, so once I get into 25 Broadway I'd
>peer with them.Mind if I ask how you got the traffic stats for Akamai?
A while ago one of my peers at OttIX (small IX in Ottawa) added my (and
the other peers) prefixes to their Akamai config. My OttIX traffic was
no more than 5% of my total traffic after that.
-Ralph
I'm already working on NYC. For Chicago an OC3 connection to AADS would
cost more than the long-haul back to Toronto.
Interestingly, it's arguable that once you pay for the OC3 (let's say $5k,
for sake of arguement) and the AADS port (another $5k), you're at $64/meg,
and thats assuming a 95th utilization of 155.
Assuming a more real world utilization of 120 mb/s, at best, you're at
$83/meg, somewhat more expensive than what several providers are selling
at these days.
-- Alex Rubenstein, AR97, K2AHR, alex@nac.net, latency, Al Reuben --
-- Net Access Corporation, 800-NET-ME-36, http://www.nac.net --
Date: Thu, 27 Jun 2002 20:53:44 -0400 (Eastern Daylight Time)
From: Alex Rubenstein
Assuming a more real world utilization of 120 mb/s, at best,
At very best. ATM cell tax in mind, I'd expect lower.
I just checked Last-Modified on AADS' pricing page, and it looks
like it was modified within two months of today. Assuming that
prices are as current as the page "starting at $4700/mo" is
indeed steep.
That same money buys a half rack and a ton of cross-connects at
Equinix...
Eddy
The pricing is a bit steep, as is MAE pricing.
However, when initially building a network and peering infrastructure
isn't it usually expected that you be able to peer at MAE-WEST, MAE-EAST,
and AADS?
I've noticed that the trend is gradually shifting away from this, as more
and more networks choose exchanges like Equinix to form their core
exchange points.
This brings up a question I've been wondering about for some time; How
many of you with peering policies that require geographical diversity on
public peering sessions specify exchange point presence requirements?
How many of you still only have public peering at the MAEs and AADS?
Joe
"Daniel Golding" <dgolding@sockeye.com> writes:
One is, Sprint won't peer with you. I'm not even sure who you work for, but
rest assured, they will not peer with you. Time spent on this might be
better utilized reading some of Bill Norton's excellent intro to peering
Dan, if you are a peer of sprint and I use the word peer as in:
1 : one that is of equal standing with another : EQUAL; especially :
one belonging to the same societal group especially based on age,
grade, or status
then I am sure things can happen.
Keeping the above definition of peering in mind, and not the current
accepted definition of what "peering" means, things suddenly become
crystal clear.
vijay "time to put back the peer in peering" gill
Dan, if you are a peer of sprint and I use the word peer as in:
>
>1 : one that is of equal standing with another : EQUAL; especially :
>one belonging to the same societal group especially based on age,
>grade, or status
>
>then I am sure things can happen.
>
>Keeping the above definition of peering in mind, and not the current
>accepted definition of what "peering" means, things suddenly become
>crystal clear.
Not trying to start a "peering" debate, but I do believe there are benefits to peering with networks which are not your "equal", in both directions.
OTOH, some networks who peered with anyone and everyone did not survive. While some networks who peered with no one have also died. (And some who peer with no one just over-report EBITDA by more than the GNP of many countries. 
So I am not sure there is any strong evidence that peering or not is good for long term economic viability.
I do believe there is operational evidence that a more open peering policy can reduce latency to off-net locations, but I am sure there are other reasons to close your peering policy.
Fortunately, there are few regulations, so most networks may peer with whomever they please - equal or otherwise.
>vijay "time to put back the peer in peering" gill
And to think you used to work for Above.Net....
"Daniel Golding" <dgolding@sockeye.com> writes:
> One is, Sprint won't peer with you. I'm not even sure who you work for, but
> rest assured, they will not peer with you. Time spent on this might be
> better utilized reading some of Bill Norton's excellent intro to peeringDan, if you are a peer of sprint and I use the word peer as in:
1 : one that is of equal standing with another : EQUAL; especially :
one belonging to the same societal group especially based on age,
grade, or statusthen I am sure things can happen.
Yeah, right. This is complete nonsense.
Sprint's peers aren't equal to Sprint or each other when considered by
revenue, profitability, number of customers, or geographical coverage.
If you mean equal as in within an order a magnitude of 10 larger or
smaller, then you cover most of their peers.
Keeping the above definition of peering in mind, and not the current
accepted definition of what "peering" means, things suddenly become
crystal clear.
This is ridiculous elitism.
You don't need to pretend that all of Sprint's peers are "equal" in order
to make sense of their peering policy, just accept the fact that each
company has policies that they believe to be in their best interests and
omit the pretense of justifying this by the movement of heavenly bodies in
the spheres.
While you are at it look up the definition of equal.
Mike.
+------------------- H U R R I C A N E - E L E C T R I C -------------------+
Mike Leber <mleber@he.net> writes:
Sprint's peers aren't equal to Sprint or each other when considered by
revenue, profitability, number of customers, or geographical coverage.
A good proxy for the above is to ask the question:
Do X and Y feel they derive equal value (for some value of equal) by
interconnecting with each other?
If they think they do, then an interconnection is set up between X
and Y. However, if one party feels that they do NOT derive equal
value by interconnecting with the other, than that party usually
balks.
X states that they would only feel equal value is derived by both
parties if traffic between X and the other party is n mb/s with a
ratio of p:q. Y disagrees. They do not interconnect. This causes
pain.
Usually the pain for one party is greater than the pain for the
other, unless they are really peers of each other, in which case
settlement free interconnections happen. However, if there isn't
equal amounts of pain being felt on both sides, then normally the
party with the more hurt tries to redress the issue.
Usually this imbalance in perceived value is redressed by one of the
parties offering to make up the difference by some form of a transfer
of money.
This is ridiculous elitism.
see above.
to make sense of their peering policy, just accept the fact that each
company has policies that they believe to be in their best interests and
omit the pretense of justifying this by the movement of heavenly bodies in
the spheres.
I think we are in agreement here.
/vijay
Usually the pain for one party is greater than the pain for the
other, unless they are really peers of each other, in which case
settlement free interconnections happen. However, if there isn't
equal amounts of pain being felt on both sides, then normally the
party with the more hurt tries to redress the issue.Usually this imbalance in perceived value is redressed by one of the
parties offering to make up the difference by some form of a transfer
of money.
and yet, the party who experiences the pain will normally perceive the
other party's *intentions* as the cause of that pain. knowing that the
pain can be transformed from "can't exchange traffic" pain into "must
pay money" pain tends to reinforce this perception.
when this situation has existed in other industries, gov't intervention
has always resulted. even when the scope is international. i've not
been able to puzzle out the reason why the world's gov'ts have not
stepped in with some basic interconnection requirements for IP carriers.