Mailing List Archive

Paul A Vixie <paul@vix.com> wrote:

>Last time this term came up, I opined that there was no "backbone" any more
>and that 1996's Internet had a "hairball topology." Vadim, among others,
>disagreed with me but we didn't pursue the topic. Perhaps we should have.

Well, "backbone" is too vague. I rather prefer to think of Internet
topology as of "tiers". The nodes in upper-levels cast "cones of
influence" in lower tiers. Nodes from lower tiers belonging to
different cones of influence do not generally speak to each other,
and so have to purchase transit from higher tiers.

Note that this loose definition provides different idea of "backbone"
aka the first tier, quite unlike the "defaultless core". To illustrate
it let's assume there are two regional dual-homed ISPs, A and B,
connected to, say ANS and MCI:

ANS- --MCI
| X |
A--- ---B

Both A and B generally have to be defaultless. But they are not
in the first tier, but rather in second tier, as they have to purchase
transit from first-tier providers to talk to each other.

May be we should classify ISPs by miles*bandwidth of their internode
circuits?

>And in that sense, there is no backbone in 1996.

Tier one is the "backbone". Those are providers not purchasing
transit from anybody else.

>Terminologically speaking, there's no discrete set of wires or routers or
>companies you can point to and say, "there, that right there, that is the
>Internet Backbone."

Well, you can do that in practice, and be pretty certain.

>We tend to reserve the term "NSP" for folks who peer at enough NAPs that they
>have no default route and aren't buying transit from anybody.

Ok, so here we agree :)

>We tend to use
>the term "ISP" when we mean someone in the packet or even the session business
>who _does_ have to buy transit from somebody. Once in a while I hear the term
>"backbone provider" used synonomously with "NSP" (as defined above).

That's pretty common usage. Sounds much better than an-as-pee.
The term for second-tier is "regional provider" and third tier is
usually local providers.

>I am not even going to get started (here and now, at least) on the subject of
>peering politics/economics. I just thought I'd chime in on the definitions of
>the words we're all using.

--vadim

>From: John Curran[SMTP:jcurran@bbnplanet.com]
>
>At 10:32 AM 4/5/96, William Allen Simpson wrote:
>>> From: Avi Freedman <freedman@netaxs.com>
>>>...
>>> Everyone (of importance) agrees that in order to claim you're a
backbone
>>> you have to (now, not a year ago) be connected to at least 2 public
NAPs/MAEs
>>> and have at least one circuit that runs at DS3 or higher speed.
>>>
>>No, that is not correct.
>>
>>A US Internet "backbone" is one which connects to ALL the NAP/MAEs in
>>the US. Not just two. All of them.
>
>Bill,
>
> I'm not sure that's a viable definition. First, the number of MAE's
> seems to be increasing withou bound, and secondly, there are points
> that you don't want to connect due to their performance. Finally, is
> "connecting" considered the same as "peering"?
>
>/John


There is no clear consensus concerning which of the facilities are "NAPs",
let alone which ones an ISP should participate in to be considered a
credible National/Backbone provider. For instance, while MAE-West was not
one of the original NAPs established by NSF, it clearly qualifies in all
other respects. On the other hand, I don't think it would be considered
necessary to be at LA, Phoenix, Tucson, Dallas, etc right now to be
considered a National/Backbone provider. The criteria I've heard most
frequently is "connected to three of the NAPs, peering with at least 2
national providers at each of those NAPs". I believe this is what MCI
requires in order to establish peering with a new entity...
--
Jim Browning

At 12:35 PM 4/5/96, Jim Browning wrote:

>There is no clear consensus concerning which of the facilities are "NAPs",
>let alone which ones an ISP should participate in to be considered a
>credible National/Backbone provider. For instance, while MAE-West was not
>one of the original NAPs established by NSF, it clearly qualifies in all
>other respects. On the other hand, I don't think it would be considered
>necessary to be at LA, Phoenix, Tucson, Dallas, etc right now to be
>considered a National/Backbone provider.

Oh, I fully agree.

>The criteria I've heard most frequently is "connected to three of the
>NAPs, peering with at least 2 national providers at each of those NAPs".
>I believe this is what MCI requires in order to establish peering with a
>new entity...

Mumble. Presumably, not every national provider will require peering
with at least two others before agreeing to peer... the possibilities for
deadlock seem a tad high.

/John

Well, connecting to an Internet backbone and having your own nationalwide backbone
are two different things.

It is even extremely difficult to define what is Internet backbone. Are MCI and Sprint
connected to backbone? I think that not only they are, they also have their own
DS3 backbone.

Is company like Aimnet connected to backbone? Well, we have DS3s to PacBell NAP
and MAE West, and of course think that we are connected to backbone although
we do not have our own backbone. But we do not claim that we have Internet backbone.

Hong
Aimnet



-------- Begin Included Message -------

> At 10:32 AM 4/5/96, William Allen Simpson wrote:
> >> From: Avi Freedman <freedman@netaxs.com>
> >>...
> >> Everyone (of importance) agrees that in order to claim you're a backbone
> >> you have to (now, not a year ago) be connected to at least 2 public NAPs/MAEs
> >> and have at least one circuit that runs at DS3 or higher speed.
> >>
> >No, that is not correct.
> >
> >A US Internet "backbone" is one which connects to ALL the NAP/MAEs in
> >the US. Not just two. All of them.
>
> Bill,
>
> I'm not sure that's a viable definition. First, the number of MAE's
> seems to be increasing withou bound, and secondly, there are points
> that you don't want to connect due to their performance. Finally, is
> "connecting" considered the same as "peering"?
>
> /John
>
>
>
-------- End Included Message --------

The NSF didn't use the term "backbone," but it did use the term NSP - Network
Service Provider. The NSF definition for NSP was

- connect to hubs called priority Network Access Points,
- route and carry traffic to or from any research and education location,
and
- make the routes available to a routing arbiter

The implementation wasn't as clean as the definition.
--
Sean Donelan, Data Research Associates, Inc, St. Louis, MO
Affiliation given for identification not representation

>> From: Avi Freedman <freedman@netaxs.com>
>> And they solicit our customers, sigh.
>>
>> And someone with 2 T1s to Sprint has been saying "We are *the* Internet
>> Backbone in South Jersey".
>>
>> Everyone (of importance) agrees that in order to claim you're a backbone
>> you have to (now, not a year ago) be connected to at least 2 public NAPs/MAEs
>> and have at least one circuit that runs at DS3 or higher speed.
>>
> No, that is not correct.
>
> A US Internet "backbone" is one which connects to ALL the NAP/MAEs in
> the US. Not just two. All of them.

Perhaps you can tell me where PSI connects, good sir...
By my count it's MAE-East and CIX and nowhere else.
There are a few major backbones not at the Chicago NAP.
And none of the majors are at MAE-Chicago or any of the MAEs other
than East and West.

> Everyone else is just a "regional", of one size or another.
>
> If anyone solicits your customers saying otherwise, report them for
> false advertising to the FTC, and sue the bastards.
>
> WSimpson@UMich.edu

Avi

From: Michael Dillon[SMTP:michael@memra.com]
>
>On Fri, 5 Apr 1996, Gordon Cook wrote:
>
>> Bill sez: A US Internet "backbone" is one which connects to ALL the
>> NAP/MAEs in the US. Not just two. All of them.
>
>Why not just start calling it the Internet "core". The core of the US
>Internet no longer follows a backbone topology. The core is composed of
>the major NSP's who operate national backbones providing national transit
>and who interconnect at all or most of the public exchange points.
>
>Make sense?

Yes, as does much of what you say... :)
--
Jim Browning

>> Everyone (of importance) agrees that in order to claim you're a backbone
>> you have to (now, not a year ago) be connected to at least 2 public NAPs/MAEs
>> and have at least one circuit that runs at DS3 or higher speed.
>>
>No, that is not correct.
>A US Internet "backbone" is one which connects to ALL the NAP/MAEs in
>the US. Not just two. All of them.
>Everyone else is just a "regional", of one size or another.

Name any ISP which meets that critiera.

[Hint: who is at MAE-Chicago right now?]

Once you start doing BGP peering at T3 speeds in two geographically distinct
regions, you're playing in big leagues. There is a tier below that of BGP
peering at one location; there is the tier above it of peering at *lots*
of places rather than just a few, but IMHO once you have the multiple
peering points you can call yourself a backbone or core provider,
and I'll gladly testify to that in a deposition or in court if you
start going around suing people who use it.


-george william herbert
gherbert@crl.com

Better yet, call it the 'default-free core', which is ironically what it
is already called. :-)

- paul

At 12:02 PM 4/5/96 -0800, Michael Dillon wrote:

>On Fri, 5 Apr 1996, Gordon Cook wrote:
>
>> Bill sez: A US Internet "backbone" is one which connects to ALL the
>> NAP/MAEs in the US. Not just two. All of them.
>
>Why not just start calling it the Internet "core". The core of the US
>Internet no longer follows a backbone topology. The core is composed of
>the major NSP's who operate national backbones providing national transit
>and who interconnect at all or most of the public exchange points.
>
>Make sense?
>
>Michael Dillon Voice: +1-604-546-8022
>Memra Software Inc. Fax: +1-604-546-3049
>http://www.memra.com E-mail: michael@memra.com
>
>

On Fri, 5 Apr 1996, Gordon Cook wrote:

> Bill sez: A US Internet "backbone" is one which connects to ALL the
> NAP/MAEs in the US. Not just two. All of them.

Why not just start calling it the Internet "core". The core of the US
Internet no longer follows a backbone topology. The core is composed of
the major NSP's who operate national backbones providing national transit
and who interconnect at all or most of the public exchange points.

Make sense?

Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-546-3049
http://www.memra.com E-mail: michael@memra.com

At 01:59 PM 4/5/96 -0800, Michael Dillon wrote:

>
>The core of the US Internet, also known as the default-free core, no longer
>follows a backbone topology. The core is composed of the major NSP's who
>operate national backbones providing national transit and who interconnect
>at all or most of the public exchange points.
>

Also, it would preferable to omit 'US' from the above; the Internet
core doesn't reside solely in the United States. :-)

- paul

Last time this term came up, I opined that there was no "backbone" any more
and that 1996's Internet had a "hairball topology." Vadim, among others,
disagreed with me but we didn't pursue the topic. Perhaps we should have.

Back in the dimming prehistory of the universe, there was a "backbone" in
the sense that AS690 really mattered and if you didn't have someone sending
your routes into ANS (or NSFnet or ARPAnet, depending on the month and year)
then you could not really say that you were "on the Internet" since most
folks would not be able to reach you.

There is no such AS in 1996.

And in that sense, there is no backbone in 1996.

What we have these days is a _whole_bunch_ of AS's such that if any one of
them does not receive your routes somehow, then you aren't "on the Internet"
since a whole lot of other end hosts will not be able to reach your end hosts.
Sprint, MCI, ANS, Alternet, PSI, and AGIS come to mind. (There are others.)

Terminologically speaking, there's no discrete set of wires or routers or
companies you can point to and say, "there, that right there, that is the
Internet Backbone."

We tend to reserve the term "NSP" for folks who peer at enough NAPs that they
have no default route and aren't buying transit from anybody. We tend to use
the term "ISP" when we mean someone in the packet or even the session business
who _does_ have to buy transit from somebody. Once in a while I hear the term
"backbone provider" used synonomously with "NSP" (as defined above).

I am not even going to get started (here and now, at least) on the subject of
peering politics/economics. I just thought I'd chime in on the definitions of
the words we're all using.

On Fri, 5 Apr 1996, Paul Ferguson wrote:

> Better yet, call it the 'default-free core', which is ironically what it
> is already called. :-)

OK, how about this...

The core of the US Internet, also known as the default-free core, no longer
follows a backbone topology. The core is composed of the major NSP's who
operate national backbones providing national transit and who interconnect
at all or most of the public exchange points.

So, to determine whether a carrier is part of the core:

Are they an NSP?
Do they operate their own national backbone?
Can they provide national transit over their own network infrastructure?
Do they interconnect with other NSP's who satisfy the previous two
conditions at most of the public exchange points?

Is this better?

Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-546-3049
http://www.memra.com E-mail: michael@memra.com

> On Fri, 5 Apr 1996, Paul Ferguson wrote:
>
> > Better yet, call it the 'default-free core', which is ironically what it
> > is already called. :-)
>
> OK, how about this...
>
> The core of the US Internet, also known as the default-free core, no longer
> follows a backbone topology. The core is composed of the major NSP's who
> operate national backbones providing national transit and who interconnect
> at all or most of the public exchange points.
>
> So, to determine whether a carrier is part of the core:
>
> Are they an NSP?
> Do they operate their own national backbone?
> Can they provide national transit over their own network infrastructure?
> Do they interconnect with other NSP's who satisfy the previous two
> conditions at most of the public exchange points?

Where public exchange points == {MAE-East, MAE-West, Pennsauken, PacBell NAP,
Chicago NAP, and arguably the CIX router/cloud}.

> Is this better?
>
> Michael Dillon Voice: +1-604-546-8022
> Memra Software Inc. Fax: +1-604-546-3049
> http://www.memra.com E-mail: michael@memra.com

Avi

On Fri, 5 Apr 1996, Paul A Vixie wrote:

> Last time this term came up, I opined that there was no "backbone" any more
> and that 1996's Internet had a "hairball topology." Vadim, among others,
> disagreed with me but we didn't pursue the topic. Perhaps we should have.

This is what I have said. I think there is 'no' backbone but
there are many NSPs that have connections to the major NAPs. There are
also compaines that connect to more than two NSPs and NAPs. Lets fiqure
this out, is there no 'backbone' or is there a 'backbone'?

> And in that sense, there is no backbone in 1996.

Agreed

> We tend to reserve the term "NSP" for folks who peer at enough NAPs that they
> have no default route and aren't buying transit from anybody. We tend to use
> the term "ISP" when we mean someone in the packet or even the session business
> who _does_ have to buy transit from somebody. Once in a while I hear the term
> "backbone provider" used synonomously with "NSP" (as defined above).

I think it should be NSP.They have National Netowrks, we have
State wide netowrks. Simple.

Christian Nielsen
Vyzynz International Inc. cnielsen@vii.com,CN46,KB7HAP
Phone 801-568-0999 Fax 801-568-0953
Private Email - Christian@Nielsen.Net BOFH - cnielsen@one.dot PS :)

At 03:34 PM 4/5/96 +0800, Vadim Antonov wrote:

>Paul A Vixie <paul@vix.com> wrote:
>
>>Last time this term came up, I opined that there was no "backbone" any more
>>and that 1996's Internet had a "hairball topology." Vadim, among others,
>>disagreed with me but we didn't pursue the topic. Perhaps we should have.
>
>Well, "backbone" is too vague. I rather prefer to think of Internet
>topology as of "tiers". The nodes in upper-levels cast "cones of
>influence" in lower tiers. Nodes from lower tiers belonging to
>different cones of influence do not generally speak to each other,
>and so have to purchase transit from higher tiers.
>

Well, I kind of liked the term 'blobs' [credit to Jerry Scharf]. :-)

- paul

On Fri, 5 Apr 1996, Vadim Antonov wrote:

> Well, "backbone" is too vague. I rather prefer to think of Internet
> topology as of "tiers". The nodes in upper-levels cast "cones of
> influence" in lower tiers. Nodes from lower tiers belonging to
> different cones of influence do not generally speak to each other,
> and so have to purchase transit from higher tiers.

> Tier one is the "backbone". Those are providers not purchasing
> transit from anybody else.

If we think like an onion (shades of TinyBASIC!) then the core of the
Internet are these providers who supply transit over their own national
and international backbones and who do not need to buy transit from other
providers. The providers who form the Internet core are sometimes called
NSP's (Network Service Providers) and sometimes called Tier 1 providers

> That's pretty common usage. Sounds much better than an-as-pee.
> The term for second-tier is "regional provider" and third tier is
> usually local providers.

The next layer of the onion is the Tier 2 providers sometimes referred to
as regional providers although they may actually serve overlapping
geographical regions. These providers do not provide transit but they do
supply other providers in a lower tier.

This brings us to the Tier 3 providers commonly known as ISP's (Internet
Service Providers. These organizations may connect to Tier 2 providers
or Tier 1 providers but their distinguishing characteristic is that they
do no normally supply organizations who resell Internet access.

Tier 4 networks belong to those organizations who provide Internet access
for their own members or employees. These could be corporations, schools,
or universities who operate both internal networks and provide dialup
services that are not available to the general public. Sometimes a Tier 4
network provides access to other organizations such as a company which
supplies its subcontractors with their Internet connectivity.

Tier 5 is the end user. They may have a single PC that dials up to the
Internet or they may be sitting in front of a workstation on a corporate LAN.

Unlike an onion skin, these layers are not precise and there is some overlap
especially in Tier 2. Until recently most ISP's connected directly to
Tier 1 providers and although there are some providers who are starting
to specialize in Tier 2 services it will remain common for both Tier 1
and 3 organizations to be in that market.

This seems to explain the relationships in a way that I think the average
person or journalist could understand and still form concepts fairly
close to the reality of today's global Internet.

Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-546-3049
http://www.memra.com E-mail: michael@memra.com

On Friday, April 05, 1996 1:55 PM, Paul Ferguson[SMTP:pferguso@cisco.com] wrote:
@Better yet, call it the 'default-free core', which is ironically what it
@is already called. :-)
@
@- paul
@

Actually some of the large carriers do have a hidden default route...to the bit bucket...:-)

--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL 60563

e-mail: JimFleming@unety.net

> From: Avi Freedman[SMTP:freedman@netaxs.com]
> Sent: Saturday, April 06, 1996 1:15 AM
> To: Michael Dillon
> Cc: pferguso@cisco.com; nanog@merit.edu
> Subject: Re: the Internet Backbone
>
> > Are they an NSP?
> > Do they operate their own national backbone?
> > Can they provide national transit over their own network infrastructure?
> > Do they interconnect with other NSP's who satisfy the previous two
> > conditions at most of the public exchange points?
>
> Where public exchange points == {MAE-East, MAE-West, Pennsauken, PacBell NAP,
> Chicago NAP, and arguably the CIX router/cloud}.
>
> And NSPISP, NXIX, STIX, HKIX, D-GIX, F-GIX, CIXP, LINX, etc. etc. etc.

No, the point is that {MCI, Sprint, ANS, UUNET, PSI, AGIS} would fail the
test if you include all of the other, smaller, public exchange points.

> Barry

Avi

At 3:02 PM 4/5/96, Michael Dillon wrote:
>On Fri, 5 Apr 1996, Gordon Cook wrote:
>
>> Bill sez: A US Internet "backbone" is one which connects to ALL the
>> NAP/MAEs in the US. Not just two. All of them.
>
>Why not just start calling it the Internet "core". The core of the US
>Internet no longer follows a backbone topology. The core is composed of
>the major NSP's who operate national backbones providing national transit
>and who interconnect at all or most of the public exchange points.

Sounds like a fine definition... (although adding the qualification
of "functional" to "public exchange points" might be worthwhile...)

/John

I frequently teach routing and operations, and find this discussion of
terms to be coming up with a useful consensus to answer some very
frequently asked questions. I'd like to see the results of this discussion
spread more widely than the NANOG list.

It strikes me that the discussion is the kernel of an informational RFC
complementary to Dave Crocker's RFC1775, "To be "on" the Internet." Here,
perhaps we are defining what it means to be "in" the Internet. RFC1775
emphasized the user perspective; it would seem that a different document
emphasizing the operator perspective also would be useful.

Michael Dillon and others have put forth some good definitions. Would any
of you active contributors be interested in trying an RFC on this? I'll
step up to editing it if it would be useful.

-----
My comments on Michael's last posting:


If we think like an onion (shades of TinyBASIC!) then the core of the
Internet are these providers who supply transit over their own national
and international backbones and who do not need to buy transit from other
providers. The providers who form the Internet core are sometimes called
NSP's (Network Service Providers) and sometimes called Tier 1 providers


> That's pretty common usage. Sounds much better than an-as-pee.
> The term for second-tier is "regional provider" and third tier is
> usually local providers.

The next layer of the onion is the Tier 2 providers sometimes referred to
as regional providers although they may actually serve overlapping
geographical regions. These providers do not provide transit but they do
supply other providers in a lower tier.

HCB> I'm not sure about the part that a Tier 2 provider never provides
HCB> transit.

This brings us to the Tier 3 providers commonly known as ISP's (Internet
Service Providers. These organizations may connect to Tier 2 providers
or Tier 1 providers but their distinguishing characteristic is that they
do no normally supply organizations who resell Internet access.

Tier 4 networks belong to those organizations who provide Internet access
for their own members or employees. These could be corporations, schools,
or universities who operate both internal networks and provide dialup
services that are not available to the general public. Sometimes a Tier 4
network provides access to other organizations such as a company which
supplies its subcontractors with their Internet connectivity.

HCB> A Tier 4 might be the first level that provides proxy aggregation,
HCB> if it supports external organizations such as subcontractors.
HCB> Proxy aggregation is more likely to be at Tier 3 or higher.

Tier 5 is the end user. They may have a single PC that dials up to the
Internet or they may be sitting in front of a workstation on a corporate LAN.

Unlike an onion skin, these layers are not precise and there is some overlap
especially in Tier 2. Until recently most ISP's connected directly to
Tier 1 providers and although there are some providers who are starting
to specialize in Tier 2 services it will remain common for both Tier 1
and 3 organizations to be in that market.

This seems to explain the relationships in a way that I think the average
person or journalist could understand and still form concepts fairly
close to the reality of today's global Internet.

On Sat, 6 Apr 1996, Howard C. Berkowitz wrote:

> Michael Dillon and others have put forth some good definitions. Would any
> of you active contributors be interested in trying an RFC on this? I'll
> step up to editing it if it would be useful.

I've been fleshing out a longer explanation in my mind last evening so
I'm willing to continue with this.

> HCB> I'm not sure about the part that a Tier 2 provider never provides
> HCB> transit.

This is true. The lines of division are somewhat fuzzy. My feeling is
that the Tier 2 provider is not primarily in the business of supplying
transit but may do so incidentally to their primary operations and even
then.


Michael Dillon Voice: +1-604-546-8022
Memra Software Inc. Fax: +1-604-546-3049
http://www.memra.com E-mail: michael@memra.com

On Sat, 6 Apr 1996, John Curran wrote:

> At 3:02 PM 4/5/96, Michael Dillon wrote:
> >On Fri, 5 Apr 1996, Gordon Cook wrote:
> >
> >> Bill sez: A US Internet "backbone" is one which connects to ALL the
> >> NAP/MAEs in the US. Not just two. All of them.
> >
> >Why not just start calling it the Internet "core". The core of the US
> >Internet no longer follows a backbone topology. The core is composed of
> >the major NSP's who operate national backbones providing national transit
> >and who interconnect at all or most of the public exchange points.
>
> Sounds like a fine definition... (although adding the qualification
> of "functional" to "public exchange points" might be worthwhile...)

Does the backbone only exist in the US then? Or have previous posters
been referring to NAP's and MAE's worldwide?

I think our US -centrism is showing here guys:)

--- David Miller
----------------------------------------------------------------------------
It's *amazing* what one can accomplish when
one doesn't know what one can't do!