Hi Ray,
You give up? Don't even think about it.
"What I don't understand here is the role of the regulators in 'establishing
the peering points."
A lot of this has to do with the knee bone being connected to the thigh bone,
the thigh bone being connected to the ... PTT.... is connected to the ITU.
So much for oversimplification.
Please pardon the discontinuities in the remaining parts of this post, as I
wrote it almost entirely out of a stream of consciousness. And, while I'm
seeking exoneration in advance, most of this post treats non-European
regions of the world, although I proceed by discussing Europe, first.
-----------------
Returning to Europe...
We'll likely see more progress made, more quickly, on the European front
than elsewhere, but even in Europe it's not as straightforward as it might
seem. Peering points might also be augmented by proxy servers, and by
localized caching of content, but these will not always help (in many cases
they will not help at all) where future e-commerce, enterprise VPN, and
public IP telephony/fax requirements exist.
It's an extremely complicated situation, the degree to which varies from
nation to nation and region to region, with the most remote from the US (the
so called third world countries) and SE Asian and Pacific Rim countries
being among the most complex. What we're really talking about here is
nothing short of the convergence of two worlds.
Everyone has gotten a taste of how difficult is it to converge two sets of
protocols like those used to support voice and data. This task (melding the
minds and spirits of the IETF and the ITU-T), I sense, is going to be some
order of magnitude larger, and several orders of magnitude beyond that
more difficult in scope, to implement.
Europe is catching up to the states more so than the other regions, at this
time, but there's still a lot of work to be done, even in Europe, before
"approved" Internet peering additions are made. Since this is the Internet,
after all, nothing takes place in a vacuum. Routes are defined that take all
possibilities into account. That means that autonomous systems need to be
redefined, routers and primary root servers need to be updated, some new
ones added to include new domains, etc.
The scope of this is international, it's global, when you start to introduce
changes to primary root and Domain Name Servers. Therefore, a greater
deal of orchestration is needed than would otherwise appear obvious. And
this orchestration demands the participation of the remainder of the world, a
universe about which much of the remainder of this post is dedicated.
[[[A major digression ensues here, some would call it a sidebar, so keep
that last thought and hold on... and then there are always the new
contracts that need to be cut that allows the worlds b-b providers to
traverse those new peering points. Some of this stuff has been done with
smoke, mirrors and handshakes in the past, but I think that we're entering a
new era when diligence is going to be paid to every bit going forward.
At the same time, as the PTT's gain more influence, we will see more of a
deterministic approach to traffic management, as opposed to best effort
which has characterized the past. I'm not saying that this will be solely
caused by the PTTs, rather, that they will be behind it 100%. Which in
some ways is good, and, depending on your philosophical outlook, in more
ways could be real bad.
You've got to read below, above, AND between the lines anytime someone
pushes an entire paradigm on you solely on the merits of a single set of
attributes, like a specific religion of QoS. There's usually a greater prize
they are aspiring to behind the door. End major digression]]]
Getting back to Network Access Points and Private Peering: Two Extreme Flavors:
I've taken some license to differentiate "approved" in this context from
"expedients." The latter (expedients) could potentially wind up being only
temporary fixes, because many of the emerging private relationships which
fit this "expedient" class are vulnerable to the whims of those (the partners)
who partake in them.
Of course, there's really no such official classifications as approved and
expedient, I've only made those up to describe the two extreme possibilities.
In NA, the approveds would be the regional NAPs, the MAEs, the CIXs,
PAIX, etc.
The ones which proved to be expedients-only were some of those that went
in as NAP bypasses, some called themselves P-NAPs, only to be bought
out later on by larger players who assimilated them into their own networks.
And then, of course, you have those whose primary source of business is
through peering and colocation services, like those of ABOV, ISLD, EXDS,
and a very large list of up and comers who are now aiming their sites in this
direction, primarily colos now, with greater aspirations down the road.
Genuity tried this, as did a few others who put in "one hop" ATM networks
at one time, but were later bought out by CLECs. These are, for the most
part, and even by some of the generic names they go under, bypasses
around the greater Internet's primary network access points.
And these private peering points and NAP bypasses exist for some very
good reasons, mostly having to do with the exclusionary, dare I say
predatory, practices of "the club" which consists of the Top Six or so Tier
One backbone providers, and the established NAP-ing hosts. But the
detailed reasons behind these would be better covered in another post, at
some later time. This message, in reply to your original question you will
come to learn, is cluttered enough the way it stands.
Deciding where and how to site a peering point is but one issue, although it
is a very important one. On the surface it would appear that the other
primary inhibitors might be: access to pipes; leased line costs; dialup billing
disparities, international route settlement formulas between the backbone
providers, etc. While the foregoing certainly play their part in the underlying
issues I alluded to, there is far more than meets the eye, elsewhere, that
must be resolved before there is a sense of assuredness about the future of
the 'net.
A single statement response to your question would make very little sense if
an understanding of many of these underlying issues are not covered and
understood first. Some of these subordinate issues may not even seem
remotely related to the main issue you raised.
In order to get to the point where the location of a particular network
access point, or even a private peering point, actually matters, many other
socially and politically ingrained hurdles need to be ironed out and resolved,
first.
------
FUDnet.
The PTTs in many parts of the world regard the 'net as a major force
which threatens to destabilize their current best interests. And this holds
true both economically and politically, if you can begin to separate those
two realms from their cultural underpinnings which are equally germane.
At present, the smaller ISPs around the world are proceeding with extreme
caution in many regions, as they open up their shops for business. In many
situations the only guarantee that they have that they will not be shut down
at a moment's notice is their dependence on the continued forbearance of
authorities, which serves to substitute for an outright sense of legitimacy.
But they'll take it any way they can get it.
In other words, they are opening up their ISP shops knowing full well that
the PTT is closing one eye while not invoking carrier stipulations against
them (since carrierdom is the only officially sanctioned form of telecomms
in those countries), and they are hoping that nothing occurs to offset this
hands off approach, fragile as it may be.
This is a pretty lousy way to go forward, wouldn't you say? But take it a
step further. These are the same shops that need recognition from the 'net's
registrars for DNS provisions. And guess who the emerging players are
who will affect who gets what in their respective regions when they
become assimilated into the Internet's governance bodies?
[[Answer: The PTTs... for reasons which become clear in a moment,
below. But keep in mind who the big daddy equivalent is to the IETF is to
these PTTs, when all is said and done. You got it, the ITU-T.]]
Beyond these somewhat ominous conditions, there are many issues having
to do with vested interests on the parts of officials in all walks of
government and public service organizations. Primarily, they are involved
with keeping themselves and others in pay checks, while at the same time
keeping many of the citizens they serve in the dark, restricting how they get
bandwidth, and how much information citizens should have at their disposal
vis a vis the powers that be.
I needn't get into all of the specifics, but they stem from issues ranging as
widely apart as basic human rights to those which center on monitoring the
flows of commerce. Essentially, it's all about preserving their present
nature and balance of the current power structures.
---------
IP Telephony.
VoIP and purer forms of IP Telephony (IPTel) are always linked in the
minds of officials to the existence of ISPs, and they are viewed as a threat
to the coffers of many nations who subsidize politically motivated venues
through the proceeds of Minutes of Use (MOUs) on their PSTNs. They
are also a threat to international settlements and accounting rates, topics
we've all read much about, over the past two years.
In some nations, there are ISPs who would become Internet Telephony
Service Providers (ITSPs), but they either hesitate to do so, or they must be
extremely cautious and wary about criminal laws on the books, which
serves as a major deterrent to them preventing VoIP/IPTel from happening
anytime soon. But it happens, nonetheless, through PC-based VoIP
mechanisms, but in ways that are far inferior, and over lines that are far
less robust, than the North American norm.
And again, there is always the threat of being punished in some way or
imprisoned in the extreme case, or being driven out of business, altogether.
One can begin to use their imagination as to how some of the smaller ISPs
stay above water in these situations. Yes, corruption is another area that
plagues young businesses in this sector in some regions of the world,
overseas, just like any other sector. But this sector is especially vulnerable
due to the lack of official approvals which serves to increase the
extortionist's advantage.
Okay, so I've mentioned some really extreme cases here. Let's work
backwards towards the more rational nations now.
"My point is that the regulators will not be in the driver's seat on
establishing peering in Europe but will have a looking over the shoulder role
to play at best."
This is some real murky stuff you bring to the table, Ray. Most nations
telephone administrations belong to the UN's ITU-T body, which is
responsible not only for setting standards and settlement practices in the
PTSN/GSTN space, but more recently has agreed to do peer level work
with some of the IETF working groups on several of its (IETF') more global
initiatives.
But how do you separate the global and domestic parts of the Internet,
when the nature of its very being defies such delineations? Of course, the
answer to this question, if the 'net is to grow and expand in the way that is
has, is that you don't delineate the Internet at all (at the layers that are
important for data transiting and services). This notion [of non-delineation]
is foreign to most telephone administrations, no... make that ALL telephone
administrations. Theirs is a model that is predicated on charges for end to
end session services, and timed by the minute, or some other increment, to
boot. Such a model demands that toll booths be erected at each operating
entity's border, with gate keeping functions which are closely tied to
accounting and billing functions at the same time.
Followers and proponents of the Internet's model views these constructs as
being more costly and administratively burdensome to the provider, and in
turn to the user, than the actual costs incurred to provide the services being
rendered. And they are right. Furthermore, borders and toll booths only
slow down passage by introducing more overhead (on many layers of
provider activity, both real time and back office), which is also
counterintuitive to the flows and recent initiatives of the 'net.
Beyond these admittedly almost quaint observations there is the issue of
where network intelligence should reside. The PSTN/GSTN/PTT gang has
maintained that there should be centralized intelligence (in the network's
core, and remote data bases), and that the end points should remain
relatively ignorant, like my grandmother's 1946 black, 15-pound, lead-based,
rotary telephone set... that still works.
This has been their model for decades, and all indications are that they will
continue to take this approach (despite some minor concessions they've
made concerning edge IP devices, which, by the way, will strive to emulate
the PSTN's other attributes to the hilt under the guise of VoIP) even as
they begin to adapt to the protocols which are defined in the TCP/IP stack.
The 'nethead mob, on the other hand, says that intelligence should reside in
the edge, following the now popular regime of David Isenberg's stupid
network model, or somewhere between the edge and the end points. The
core, ideally under this model, would be both blind and stupid.
Do you think that these differences will impact the forthcoming protocols
and services that are now being defined in IETF working groups if the
PTTs have a say? Well, to some extent, the ITU in effect has already
agreed to become involved, and is involved to some extent already with
defining future domain name registries, naming and numbering of end points
and SP domains, and yes, the definitions for advanced Internet services, as
well. Is it any wonder that they've recently shown an interest lately in
having a say in the emerging VoIP/IMTC/TIPHON/etc., standards?
It is due to the latter that many of the Internet's long standing proponents
and architects are now worried that the PTTs (and even their North
American counterparts, now) will begin tampering with emerging IETF
protocols and attempt to alter the 'net's basic architectural precepts going
forward.
Begin, nothing. Changes can already be seen in the philosophies behind
many of the larger so-called multiservice Internet engines now being
produced by the NT/LU and even CSCO boxes. The ones, specifically, that
are geared to multiservice applications featuring voice, fax, and the usual
run of the mill RFCs. Note, if you have the time, check out to see how
many of these are now SS7 compliant, and which point to databases which
reside within the incumbent telcos' IN/AIN information database
repositories, which, by the way, reside in the network's core.
[IN/AIN = Information Network/Advanced Information Networking]
------------------
A New Bureaucracy Is Born
Consider the role of ICANN. ICANN stands for Independent Corporation
for Assigned Names and Numbers. ICANN replaced, or otherwise took
over by some means that I am still trying to decipher, the functions which
once belonged to IANA.
ICANN is now in the process of taking on delegates (read PTT officials
and overseas vendors, maybe some academics) to its staff from nations
from all around the world who have hitherto never participated in what
have been in 'net circles, up until now, OPEN discussion forums.
Imagine this if you can. The netheads, mostly from North America and
select parts of Europe and Asia for the most part, are now assimilating into
their governance ranks PTT and ITU bureaucrats to begin multilateral
development work towards the ends which we've discussed in the previous
posts leading up to this one. The netheads prefer openness in discussion,
using argument and distention, if warranted, as a means of coming to
consensus.
The latter has been one of the defining principles upon which the 'net has
flourished. On the other hand, the bureaucrats from the PTTs and their
ITU-T brethren refuse to even allow the minutes of the meetings to be
made publicly known, much less allow any outsiders from attending or even
listening in. This gets really really sticky, so I'll leave it open ended for you
to pursue on your own if you are interested. But one can see that these
folks are not exactly off to an auspicious beginning. And yet, the near term
improvements we're discussing, not to mention the very destiny of the 'net
itself, remains in their collective hands.
The significance of this is profound both culturally and administratively. And
since the delegates to these sessions are mainly from the PTTs, there is
linkage up the line to ITU reporting and approval processes, as well. As
motions are made and adopted at the ITU level, they must proceed in a
lockstep fashion across all member organizations, many of whom live only
for and by the rules of the bureaucracy experience, itself.
The ITU is a $250,000,000 PLUS per annum service bureau which prints
every report it makes in at least several dozen different languages, whether
anyone reads them or not. The entire saga behind the contracts associated
with this printing process would take several threads to fill. But you can
assume from what I've just said that it gets deep with bureaucracy, if
nothing else.
Oh yes, if you want to read one of these things you have to subscribe on a
paying basis for the privilege, since they have not come to regard such
publicly funded works (through UN dues etc.) worthy of free public
dissemination to the world at large. Nor even to vendors, who don't pay up.
[Note, some progress has been made in this area, since "some" of the
jointly derived (IETF-ITU) works that have recently been published are
now available on the 'net. But the majority of works which are still in
progress (the really important ones where the future is at stake), and the
major historical standards and practices of the ITU over the years (most of
which are still relevant for designers needs) are still available on a fee-only,
pay as you go, basis.]
These principles are entirely out of step, they are entirely out of
synchronization, with the principles of information sharing, and the ensuing
levels of innovation, which have made the 'net what it is today. Instead of
this speaking well for the "try it, if it's broke fix it" approach, we can expect
to see i's getting dotted, and t's getting crossed ad infinitum, where not only
international issues are at stake, but where domestic ones here in the States
are concerned, as well, since ICANN's purview is international in scope.
And to repeat, once again, when you introduce changes to the 'net through
the ratification of new RFCs, you do it for the entire 'net. There are no
delineations that protect one region's protocols, practices, and services,
from any of the others. That's the whole idea. But in order to get it to work,
the parties involved must be able to see eye to eye on process, and it
doesn't look like we're in full swing in this respect, not just yet.
While it is true that enlightenment is gradually taking hold in many of the
global regions, the means by which they are willing to partake in study
groups, and joint discussion in general, is 180 out of whack with the ways of
the original Internet Society and its current heirs who for the most part
reside here in the States. As much as they say that they want to adapt, it's
not in their blood to do so yet. It's going to take some time, if not a
transfusion or two, in order to get the processes up to speed.
"The recent history in the US has been that private peering, permitted
but not established by the regulators is the solution de jour."
In many cases (not necessarily the ones you've cited) I regard the private
peering fixes as expedients, as I noted previously. Some will grow into de
facto NAPs, and some will not. Some of the more celebrated ones of the
past have proved to be tenuous, at best, and some have yielded to
decomposition over time, when one or more of the primary players was
either picked up in an acquisition, or decided to change their LOB. I see no
reasons why the same variances of chance should not be manifest in other
parts of the world, as well.
This may very well be one of the more interesting parts about the 'net to
contemplate, going forward: Peering Point Predictability. When all routes
are accessible on relatively equal terms due to the abundance of fiber
overlays, which nodal points will survive the peering glut? Now, there's one
for you to think about, eh?
This doesn't seem to bother anyone, but from an architectural view, and
from the standpoint of building extremely expensive high capacity pipes that
take months and sometimes years to get installed, I view dealing with the
smaller peering entities as a rather risky form of business. But life has to
start somewhere, right? In fairness, when private peering works out well,
it's a super thing to have.
Well, it's getting late for me. Somewhere along the line I digressed too far, I
suppose, from your original question. But most or all of these digressions, I
believe, constitute the building blocks of information that I think are
necessary to begin to see the magnitude of the problem which you
presented.
I hope I answered at least part of your question, and I would welcome
corrections and comments, and additional explanations from others as to
what goes into, and how does regulation affect, the creation of new peering
points around the globe.
Regards, Frank Coluccio |
