[NANOG]Re: Proposal for HENETIX: A Global IXP with High-Discounted Peering and Transit Services

[Matthew Petach <mpetach () netflight com>]

On Sun, Mar 2, 2025 at 9:58 AM Volkan SALiH <volkan.salih.06 () gmail com>
wrote:

> *Subject:* Proposal for HENETIX: A Global IXP with High-Discounted
> Peering and Transit Services
>
> *Dear Mr. Mike Leber, and interested NANOG members
> *

Ooh, I think that means me!  (interested NANOG member, that is--I am not
now, nor have I ever been Mike Leber)


> I hope this email finds you well. I am writing to you regarding the
> recent pricing structure adjustment for BGP tunnelbroker services, which
> were previously offered as a free (feeless) solution. Given the
> increasing demand for cost-effective and scalable interconnectivity
> solutions, I would like to propose the establishment of *HENETIX*, a
> global Internet Exchange Point (IXP) designed to facilitate
> high-discounted peering and transit services across a worldwide network.

Odd; when I go to tunnelbroker.net, I don't see any mention of a fee
increase mentioned.
But let's just put that part aside for a moment...


>       The Concept of HENETIX
>
> HENETIX would operate as a global IXP, providing Layer 2 transport
> services without additional per-port or per-transport costs, allowing
> participants to exchange traffic with remote peers efficiently. Unlike
> traditional IXPs that impose substantial fees for remote peering or rely
> on third-party transport providers, HENETIX would integrate transport
> services within a unified pricing model. This approach would align with
> industry trends favoring cost efficiency, network expansion, and
> improved global routing performance.

"unified pricing model" -- by which I would assume you mean "I want to
pay the same price to have my bits carried through
MPLS/SR/EVPN/LatestL2MagicProtocol
tunnels to anywhere on the planet as I would if I handed them to you as
IP packets across an ordinary transit port".

In other words, "I would like you to do a great deal of work, add
complexity
to your network and thus increase the fragility of the overall system,
without
earning any additional revenue to offset those negatives."


      Supporting Global Connectivity: Comparable Industry Examples
> Several networks and IXPs around the world have adopted similar models
> to reduce costs and improve accessibility. Some notable examples include:
>
>   * *PacketFabric* and *Megaport*: Both platforms offer global Layer 2
>     connectivity with a simplified pricing model that integrates
>     transport within the overall service, enabling customers to reach
>     multiple regions without complex fee structures.
>   * *Equinix Fabric*: Provides global interconnection across multiple
>     markets with predictable pricing, reducing the barriers for
>     enterprises and ISPs to participate in global exchanges.
>   * *DE-CIX, AMS-IX, and LINX*: While these IXPs operate within
>     traditional frameworks, they have introduced remote peering models
>     that enable global reach with relatively lower costs compared to
>     traditional carrier-based transport.

As you point out, there's already other players doing this; what you don't
mention is that none of those players use a "unified pricing model"
"without additional per-port or per-transport costs".  Instead, all of the
existing
players that provide an L2 connectivity service do so at a higher cost than
simple traditional IP transit, because they all recognize that carrying
traffic
through tunnels brings with it additional work and complexity that needs to
be paid for somehow.

To put it in perspective, looking at the first company you mentioned,
currently
packetfabric's pricing documentation (
https://docs.packetfabric.com/billing/services/ports/)
mentions that their 1Gbps port price is $250/month.
Doing a quick google search for "he.net transit port cost", the first
result is the sponsored link from he.net that says
"10 Gbps IP Transit $550/month | 40 Gbps IP Transit $2000/month"

So, doing the math, 1Gbps for $250 versus 10Gbps for $550 means the
competition
is successfully charging 500% more for tunnelling traffic at layer 2 than
the comparable
Layer 3 transit port would cost.

What you're proposing is that he.net should offer that same tunneled L2
connectivity
at $550 for a 10Gb port, undercutting the competition by 80% while
increasing the amount
of work they need to do.


      Legal and Regulatory Considerations
> The need for affordable interconnection services is not only a market
> demand but also a recognized regulatory priority in many regions.
> Organizations such as the *FCC (Federal Communications Commission, USA)*
> and the *European Commission* have emphasized the importance of neutral
> and cost-effective peering points to foster competition and improve
> internet resilience. Additionally, policies encouraging open
> interconnection are being promoted by various regulatory bodies to
> prevent monopolistic pricing practices in the transit market.

But nowhere in the world are regulatory agencies saying that you cannot
charge more money for a service that requires increased complexity.
Inherently, peering points are recognized to be single-location fabrics;
hence the "point" in the name "peering point".  Nobody is advocating
for an increase in the number of "peering clouds" for the simple reason
that they don't really help the internet overall.  The only recognized
benefit to converting IP packets into tunneled frames/cells/magicJuju
to carry them around the planet where they are then extracted back
out into their native IP packet format in order to be handed off to
someone else is because you want some illusion of data privacy
and address separation and isolation.  If you're simply tossing the
L3 packets back out into the wide and wooly internet afterwards,
going through the extra work of layer-2-izing them first gained you
nothing beyond some extra clock cycles spent in hardware at each
end.
Which is largely why no regulatory body is putting forth any mandates
or language about requiring increased numbers of peering clouds; they
really don't make much sense to implement.



>       Benefits of HENETIX
>
> A HENETIX platform would provide significant advantages to both
> large-scale and emerging network operators:

I notice "he.net" is notably absent from that short list of entities that
would
purportedly benefit from this proposed platform...


>  1. *Cost Reduction*: A unified membership model would eliminate high
>     transport fees, making global interconnection more affordable.

Right--so, customers would pay less, which would reduce the revenue
potential for HE.  Got it.


>  2. *Enhanced Network Performance*: By reducing dependency on Tier-1
>     transit providers, participating networks would achieve better
>     latency, redundancy, and overall performance.

No.  Just...no.  Nowhere has it been demonstrated that getting a layer 3
packet from point A to point B takes less time if you convert it to layer 2
first, then transport it, then convert it back to layer 3 at the far end,
versus
simply carrying it as layer 3 all the way along.   Encapsulating packets
does
not somehow increase the speed of light through fiber, nor does it
materially
decrease the time spent transiting silicon as it moves through intervening
devices (cue someone popping up to talk about destination port lookup times
for short labels versus IP addresses on line cards).  Your packets will not
get
to the far end materially faster just because you spent time encapsulating
and decapsulating them at each end.


>  3. *Simplified Peering*: A single interconnection framework would allow
>     participants to seamlessly peer with networks worldwide without
>     dealing with individual transport providers.

Right--so, it's simpler for the customer, but way more complex for HE.
And as noted earlier in your proposal, you're not willing to pay any
additional
amount to cover the costs of that increased complexity that he.net would
need to incorporate into their network to do this.


>  4. *Increased Market Accessibility*: Smaller ISPs, data centers, and
>     cloud providers would be able to join a global interconnection
>     fabric without financial constraints.

But they would only do so by being able to shift their costs onto he.net,
without being willing to compensate he.net for taking on those costs.


>  5. *Scalability*: As internet traffic demands continue to grow, a
>     streamlined interconnection model would ensure scalability without
>     prohibitive infrastructure investments.

This is flat-out wrong.  Scaling an L3-to-L2 encapsulation system takes
more hardware, not less.  Sure, it would allow you, the customer to scale
up "without prohibitive infrastructure investments"; but that would be true
only because you wanted your provider to bear the cost of those
infrastructure
investments without being willing to shoulder some of the costs of those
investments.


>       Proposal and Next Steps
>
> Given HE.NET's extensive network infrastructure and existing peering
> relationships, establishing HENETIX as a global IXP would be a strategic
> extension of your current services. By leveraging your existing assets
> and expertise, this initiative could position HE.NET as a leader in
> cost-effective global interconnection solutions.
>
> I would appreciate the opportunity to discuss this proposal further and
> explore potential strategies for implementation. Please let me know a
> convenient time for a meeting or call to delve deeper into this concept.
>
> Thank you for your time and consideration. I look forward to your
> thoughts on this matter.

To sum up (paraphrasing, of course):
"I would like you to embrace a more complicated and less reliable model for
carrying IP traffic around the planet that will increase your costs, but I
would
like for you to do so without charging me any additional money to do so."

I would like to wrap this up by pointing out that he.net already provides
layer 2 transport as an option: https://he.net/layer2/
You won't get it for the same price as your IP transit port, but you can
already
build your own layer 2 links to remote peering locations across their
network.
So, you can go ahead and build HENETIX yourself today, (though I might
recommend
choosing a different name for it), no need to wait for Mike to finish
facepalming first.

Sincerely,

Matt
"interested NANOG member", not a hurricane electric employee
_______________________________________________
NANOG mailing list 
https://lists.nanog.org/archives/list/nanog () lists nanog org/message/WAQAHAQQNXHWL5POBEMXIF7CCY2TEVLW/