5/98 Boardwatch/Rickard article (Part II)
boardwatch.com
ISPCON RAVINGS
I announced at the opening session of the Internet Service Provider Convention
that I did not believe the size of the proposed WorldCom/MCI merger was at
all the problem and rather surprised everyone in coming out in favor of the
merger. Rather, that the extant non-policy peering policy was the biggest threat
to the future of a competitive Internet. This rattled a few cages. But I still think
it's true. If we consider that the future of a competitive Internet rests on whether
or not WorldCom/MCI merge, then we must logically conclude also that other
future mergers could similarly threaten a competitive future for the Internet. And
if the solution is for the government to review and disapprove this merger, then
it pretty much makes sense that to guard a competitive environment, it must also
review and approve or disapprove EVERY merger or acquisition of any
companies involved in the Internet on a case-by-case basis. I don't think
Martha Stewart would find this to be a "good thing." In fact, without even
consulting Martha on the topic, I find it to be potentially a very, very bad thing.
And that brings us back to the original problem. How to have a competitive
Internet where new players can enter and smaller players can innovate and
introduce new products without filleting large networks that have invested
hundreds of millions of dollars in their networks. One way NOT to do it is with
the current structure where networks show up at whatever NAPs they want,
with whatever bandwidth they want, and peer with whoever they want - thus
designating who may BE their competitors, under what conditions, and at what
bandwidth. It is time to admit in public that this may have got us here, but it just
didn't work long term. It happened due to a lack of intestinal fortitude and/or
political courage in 1994. It will require a good bit of both to fix in 1998, but it
really must be fixed.
I would advocate a radical new peering policy with several major features:
1. Compelled Peering. Exchange participants would be required to
peer with similar class networks at the exchange. It would be quite
mandatory and quite automatic. Forget costs of "enabling your
competitor." They're a cost of being connected to the Internet. If you
don't want to pay these largely imagined costs, run a private network for
your own entertainment and that of your friends. Don't be surprised if the
group of friends turns out to be smaller than you thought. If you want to
be connected to the Internet, get onboard this new peering era.
2. Required Bandwidth. Throwing a symbolic T-1 into CIX is not
participating in an Internet, nor is it peering. Exchange participants should
be required to bring enough bandwidth to handle 120 percent of their
traffic at all times - peak - at all exchanges through strict monitoring and
rules enforcement.
3. Expanded Number of Exchanges. If voice, fax, and ultimately
video are to move onto IP networks, it does not make sense to run all
that traffic through four rooms in the United States (or am I alone in
thinking this is madness?). Similarly, exchange should not be a secret,
accidental network design among those colluding to exclude other
players. There are a number of ways to accomplish this. I personally
favor requiring all "peers" at the exchanges to also PROVIDE a public
exchange on their network with collocation facilities, etc. In this way, as
more backbones are added, more exchanges are added as well.
WorldCom could start with about four - one each for UUNET, ANS,
CompuServe, and GridNet. We should have a competitive Internet - not
a free one where anyone with a dial-up modem can be a backbone. As
traffic rises and the Internet must grow, national backbone status is going
to have an escalating price in appearing at an ever larger number of
exchanges. But there would be room for "regional" networks and other
classes of "peers" in a rational order participating at smaller numbers of
exchanges.
4. International Peering. What's going on here? Foreign networks
backhaul all the way across the ocean to the United States, and THEN
have to pay transit to the networks here under the dubious theory that
the U.S. network is hauling their water across the continent? Good work
if you can get it. But it's an international screw job. Long term we face
the same instability. What if they tire of this and disconnect? Four years
down the road and it will be the same confrontation, over which
CONTINENT has the most customers, the best customers, yata, yata,
yata. Deal with this now.
5. Governing Body. If you don't want government regulation, you do
need a persuasive form of self- regulation to forestall it. And despite my
simple/radical view of peering, there are ongoing issues of practical
nature in peering and traffic exchange that don't solve themselves in
magazine articles. A real, neutral, governing body must be established to
manage the exchange of traffic and route advertisements and deal with
the "Jimmy peed in Sally's lunchbox and I'm telling mom" type of issues.
The choice here isn't between no regulation and a governing body. It's
between an industry body and a government body. For any of those who
missed the study group, the correct answer is A.
So I would advocate that rather than stop the merger, let's fix the real problem.
Fortunately or unfortunately, the merger question provides us a unique moment
in time - a huge opportunity. It's a big merger, with big players. If it were "held
hostage" to a new peering policy that guarantees a competitive Internet, then
once that is resolved, the merger has no teeth to bite the rest of us, and must be
viewed as a pooling of interests to gain certain efficiencies in the marketplace -
good sport all around. Let's trade a merger for a long-term viable competitive
peering and exchange policy.
PEERING IS CONTROVERSIAL
So there is my take on competition, the merger, and peering. Fortunately or
unfortunately, there are a little less than 100 humanoids on the planet whose
opinion on peering counts, and guess who ISN'T on the list. Me. And, in fact, a
quick survey of all 100 would indicate that less than 0.001 percent of them
(some guy's car actually) agree with me on peering at all. Worse, they EACH
would seem to have a plan, and all 100 plans are pretty much mutually
exclusive as best as I can tell. Herding backbone providers, predictably
enough, is again very similar to herding cats with a four iron in a lightning storm.
First it's dangerous as hell, you could be hit by lightning. Second, it's noisy as
hell, lightning storms always are. And finally, it annoys the shit out of all the cats.
Getting them to all actually go in one direction, even if it is to ensure their own
survival, is almost impossible.
But one plan did catch my eye. Mike Gaddis of SAVVIS Communications has
authored a white paper titled Brokered Private Peering that hits most of my
shopping list at least a glancing blow - a little weak on the NAP expansion, but
it leaves the door open there and nothing notable on the International issue. But
all in all a very good start. It also ameliorates the local loop costs of appearing
at multiple NAPs in an interesting way.
Gaddis, and indeed SAVVIS, are a little bit of an area of fascination for me
because they inherently think outside the box and seem to find ways of making
things work when they don't work. They showed up on the scene with this
wacky thing about NAPs nearly two years ago. They created private NAPS
and actually bought connectivity to the big three at these locations and don't
appear at public exchanges and don't peer at all. I dismissed this as
"checkbook peering" initially. But after looking at it, and after noting some
superb scores on our backbone performance indexes by SAVVIS, I've
become a believer. They have, if nothing else, shown one way toward scaleable
performance - an expensive way at this point, but a way.
Politically, they're not in the middle of anything. Nobody even knows much
about them. And Gaddis' plan is one of the best thought out we've seen - even
in draft form. We are publishing it in its entirety in this issue. Note that it is a
very rough draft, very preliminary, and discussions were going on with virtually
every major player in the business at our press time.
It's not the first plan, but it does appear to be more sweeping and
comprehensive than most. PSINet actually radically altered their own peering
policies and announced free peering last fall at ISPCON'97 in San Francisco
and were again making hay with it in Baltimore this spring. They had recently
gained a lot of OC-48 bandwidth in an equity deal with ICX. They've signed up
nearly 125 ISPs with their "free peering" concept but it is somewhat unilateral
and inevitably leads to a transit purchase from PSI. According to PSINet Vice
President John Kraft, PSI is interested in the BPP Group proposal but was not
going to act as a leader on this initiative. They want to see what happens with it.
Not to be outdone, AGIS's Phil Lawlor was quick to point out they have a plan
too, and we'll be able to see it as soon as they get it typed up. According to
Lawlor, unlike the SAVVIS plan, which is based on ATM, theirs addresses
access by EVERYTHING with the lone exception of perhaps FDDI. I almost
wish this one would emerge on top just so I could fly out and SEE a network
access point exchange room designed by Phil Lawlor.
Similarly, according to Gene Noble, vice president of Business Connectivity
Solutions at DIGEX, they are also implementing major revisions to their peering
approach to ensure the best connectivity for their customers. They had
reviewed the BPP plan and were indeed interested in how it developed, but
were not willing to jump onboard the bandwagon at this early stage.
It seems everyone agrees peering needs changes, and they unanimously agree
that no one is going to agree on anything. Everyone indicates THEY have a
peering plan, though most are not precisely specific enough to be written down,
or have much detail.
It's a classic Internet situation - everyone milling around lost, straining to find out
which way the wind blows so they can get out in front and appear to have
invented it first, and desperate not to appear to be agreeing with anyone else in
the meantime. Some days it must feel pretty good to be a John Sidgmore or
Bernard Ebbers.
But behind the scenes things are not so dark. A number of new players have
come into play or are trying to. Large, but almost silent networks such as
AT&T and IBM are increasingly disenchanted by the status quo and the
emerging WorldCom/MCI new world order. Infrastructure companies such as
Qwest Communications, L3 Communications, and Williams
Telecommunications are also wanting to be players as well as virtually all of the
regional Bell operating companies. The cobbled together NAP architecture has
never appealed to these companies who are accustomed to providing
carrier-grade quality assurances to customers who do not understand why they
now can't when it comes to Internet services. Williams Telecommunications has
come out publicly to be among the "founders" of the BPP Group and both
Qwest and AT&T are rumored to be interested. Small backbones such as
Exodus Communications and Electric Lightwave have similarly signed on. And
Ascend Communications may have stolen a march on the other hardware
manufacturers by signing on as an early sponsor of the effort with technical
assistance and their Ascend GX550 25-100 Gbps ATM Switch.
One of the drivers behind all of this actually has not much to do with peering
per se. There are a host of new Quality of Service products waiting in the wings
that simply won't work across the existing NAP structure. Everyone is excited
about doing voice over IP, for example, but the existing level of quality of voice
across the Internet is unlikely to appeal to anyone but experimenters and very
strong power users. And there is no way for any one network to unilaterally fix
this. Provisions for these sorts of services have to be provisioned in the
interconnect space between networks or they just never can quite work.
In our estimation the Gaddis white paper presents one of the best blueprints
we've seen for a radical new peering policy for everyone across the network
that would probably work. Now if we can just get EVERY network to come
out with their own different plan, we can all finally agree on whatever Bernard
Ebbers and WorldCom tells us to agree on.
The least we could do is offer a private Peering Summit Meeting (/golf
tournament, of course) here in Denver for late May or early June. This would
be small, by invitation only, with backbone representatives, a few FCC
observers to hover and look concerned, some hardware guys to swear that
whatever is agreed could in fact be done, and perhaps some caddies.
Jack Rickard Editor Rotundus
Editor: Jack Rickard - Volume XI: Issue 5 - ISSN:1054-2760 - May 1998
Copyright 1998 Jack Rickard - ALL RIGHTS RESERVED
Fable Of Contents |
