Events Add an event Speakers Talks Collections
 
SIGCOMM 2020
August 11, 2020, Online, New York, NY, USA
SIGCOMM 2020
Request Q&A
SIGCOMM 2020
From the conference
SIGCOMM 2020
Request Q&A
Video
A Public Option for the Core
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Add to favorites
137
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About the talk

This paper is focused not on the Internet architecture - as defined by layering, the narrow waist of IP, and other core design principles - but on the Internet infrastructure, as embodied in the technologies and organizations that provide Internet service. In this paper we discuss both the challenges and the opportunities that make this an auspicious time to revisit how we might best structure the Internet's infrastructure. Currently, the tasks of transit-between-domains and last-mile-delivery are jointly handled by a set of ISPs who interconnect through BGP. In this paper we propose cleanly separating these two tasks. For transit, we propose the creation of a "public option" for the Internet's core backbone. This public option core, which complements rather than replaces the backbones used by large-scale ISPs, would (i) run an open market for backbone bandwidth so it could leverage links offered by third-parties, and (ii) structure its terms-of-service to enforce network neutrality so as to encourage competition and reduce the advantage of large incumbents.

About speakers

Yotam Harchol
Research Scientist at DFINITY
Dirk Bergemann
Professor at Yale University
Nick Feamster
Director of Center for Data and Computing at University of Chicago
Eric Friedman
Co-Founder & CTO at Fitbit

Yotam Harchol is a postdoctoral researcher with VMware Research. He completed his PhD studies at the Hebrew University of Jerusalem, Israel. His research is focused on improved performance and security of network functions and middleboxes.

View the profile

Dirk Bergemann is Douglass and Marion Campbell Professor of Economics at Yale University. He has secondary appointments as Professor of Computer Science in the School of Engineering and Professor of Finance in the School of Management. He received his Ph.D. in Economics from the University of Pennsylvania in 1994. He joined Yale in 1995 as an assistant professor, having previously served as a faculty member at Princeton University. He has been affiliated with the Cowles Foundation for Research in Economics at Yale since 1996 and a fellow of the Econometric Society since 2007. Dirk Bergemann was Chair of the Department of Economics from 2013-2019 and Co-Editor of Econometrica from 2014-2018. He was recently appointed as Co-Editor of American Economic Review: Insights, 2020- and is a Member of the Executive Committee of the Econometric Society 2021-. His research is in the area of game theory, contract theory, venture capital and market design. His most recent work is in the area of dynamic mechanism design and dynamic pricing, robust mechanism design, and information design. His research has been supported by grants from the National Science Foundation, the Alfred P. Sloan Research Fellowship, Google Faculty Fellow, the Knight Foundation, the Omidyar Network and the German National Science Foundation.

View the profile

Nick Feamster is Neubauer Professor of Computer Science in the Department of Computer Science and Director of the Center for Data and Computation at the University of Chicago. His research focuses on network security and performance. He received his Ph.D. in Computer Science from MIT and his S.B. and M.Eng. degrees in Electrical Engineering and CS at MIT.

View the profile

Eric is our Co-Founder and has served as a member of our board of directors since March 2007, and most recently as our CTO. Previously, Eric served as an Engineer Manager at CNET Networks, as a Co-Founder of Wind-Up Labs, a founding engineer of Epesi Technologies, and a technical member of the Real-Time Collaboration Group at Microsoft Corporation. Eric holds a B.S. and M.S. in Computer Science from Yale University. He was selected to serve as a member of our board of directors due to the perspective and experience he brings as our Co-Founder and CTO.

View the profile
Share

Hello, my name is Scott Shaker and on behalf of my esteem. The co-authors I will be talking about public option for the core. Start off by making some distinctions, there's technology in their ways. Best effort that came out of a lot of deep thought. And we've spent a lot of time discussing Clean Slate, alternative focusing on infrastructure which is how the various entities arrange themselves, provide end-to-end internet service. This is the domain structure, how they peer, how you provide Transit contracts, the architecture. The infrastructure was

not carefully plan, but is more an accident of history and has many problems. But we, as a community who spent little time, talking about alternative the paper presents one, such alternative. And in this talk, I will motivated by discussing three problems. The first is the privatization of Transit today. In contrast to the Past most traffic, go straight from the origin to Maine to the destination to me without any Transit on the public, that doesn't mean the back of bandwidth is something used but it is mostly moved into private networks. These private networks aren't a bad

thing, they enhance customer experience. But the question is with these private networks, flourishing will the internet language, will it be left behind? Or will it still be able to provide high-quality low-cost transit to all comers? The second problem is transitive. The classical. Relationships with customer, provider Pier, or transitive. That is my providers provider apps. Like my provider in pgp itself, is transitive in the strongly affect her money flows. But the full value is more complicated than that

value. Is created consumed edges, in terms of content and dads and it's fair to point out in their seminal paper that you think of this, in terms of eyeball networks in content that works. The problem is that this lack of alignment between the flow of money, in the flow value leads to policy tussles. The most well-known policy tussle is that a network neutrality which is our third problem. That will keep following is a huge subject, but Henry has really crystallized it by observing the network neutrality is about money, not packet. And if it's about

money we should look that you can Onix Mega nomics in network neutrality. Far from settled, the literature has many models, many conclusions until there's very little Clarity coming out of this literature at the same time. The enforcement of network neutrality is a nightmare that is obviously a global infrastructure. But Network neutrality is controlled by a patchwork of local regulations to what we need is a policy Clarity that will tell us what's the right policy and we need enforcement of consistency. Once we decided on a call, see how do we

enforce it consistently in This Global infrastructure. To address these problems. We propose a public option for the core of POC. It is nothing more than a global nonprofit Transit infrastructure initially, we imagined it based on least bandwidth so that there's not a huge Capital expense in the beginning. And why smile providers Are All One Piece would attach to it and use the pot for Transit. These could either be or actors in the ecosystem things like content service providers who did

tattoo the directly to the park or to the LNP and and customers who would typically attached to a last-mile provider. Now park changes, the nature of the infrastructure from traditional that with a much simpler flow wear for the pocket of the directly attached lnp's, there's no bgp. There's no transitive. Instead you go from the source to the park to the destination at 1 p, Centerpoint Parkway. Imagine that is deployed alongside the current Internet. That is the park would be connected to one or more traditional eyes, and nothing changes. For those who don't connect to

the park, the rest of the internet, Tupac to The Pierces, another eyes internet overnight. Instead it's incrementally Deployable proposal. To what benefits does it provide? What are the dresses and free problems that I discussed earlier. It makes high quality Transit available to all Last Mile providers and reduces the advantage of those with large private Networks. It separates access from transit, actresses wear value, arises and consumed, or costs are huge present by itself has no

value by separating. Those two, it brings the flow of value, in the flow of money into better alignment. Wesley, networking travel regulations, can be enforced through terms of service so there's no need for international agreements. So there are two technical challenges that we discussed in the paper. The first is, how do you compensate bandwidth providers? That is the people who are providing these police blanks. You want to minimize the Strategic Behavioral Health Providers. So they don't spend all the time trying to outsmart

each other. And the answer here is a strategy proof option that we described in great detail in the paper. The second technical challenge is how to model the economic impact of network neutrality. And then involves first Examity what network neutrality would mean in the context of the pot and then analyzing whether would be a good thing or bad thing in this setting. To our findings here. We focus first on a very simple notion to network neutrality that of termination fees. A termination fee is where I lost my provider will go to some content provider and say you must pay me

to reach my customers even though you're not directly connected to me, I'm not going to let your package reach my customers, unless you pay me directly. So obviously gross. But it's simple to analyze and what we can find is any level of termination fees, reduces social welfare than just the overall welfare of society. We also find a termination fees, favor, and cumbent l m, p z and content providers over new entrants and increases competition. So, given those benefits, you have to ask who the pot become a reality in here.

The large sea is peas, who have these very large private Network, they could collectively provide an initial power pilot. And you could see with that pilot how competitive it is with commercial Alternatives and if it starts attracting Olympia's, you can take off from there. He might ask, why would the sea is peace to this? What would motivate them in the answers to word, fear, and greed fear? Because the lack of network, neutrality is a real threat to see ESPYs of all the sudden real problems to the

street and greed. There's an outage in business that you should come out of ties. Your compliments. That is, if you have a product that requires two components A&B, and you provide a venue on B2B commoditized because that leaves more profits left over for you. What internet service provider and the pot makes most makes both Transit and lnp's, less expensive, leaving more profit over for the content providers. So to summarize quickly, the internet infrastructure is really an accident of History wasn't carefully. Thought-out and designed to wear architecture

was, and that means it has some problems. The pot helps address these problems by one making. High-quality Transit a commodity available to all even if you don't have a large private Network. It also aligns the money flow with the value flow. Creating what we refer is the economic architecture of the internet. And it enforces Network neutrality through its terms of service. And this might be more than just a pipe dream. We understand that this is a radical proposal but it might actually happen because just like with network neutrality,

the park is more about money than about packets and the content providers have very strong incentives to making this happen. Who was that? I thank you for your attention. I put Scott. Thank you so much. Lot of great questions coming in. I'll start reading them off off of slacks or a cheese thing. From umass-amherst asks, how does the proposed economic architecture for the internet interact with pricing models from previous work? That means it's all similar challenges. With different route In fairness particular, think the congestion pricing by marrying at all customer

generated revenue and Nash. He has that directly benefits years. Quick, question pricing actually is addressing a quite different problem and it's trying to evaluate the externality, the various Flo's Place, Another flows by causing congestion and assigning prices to Flo's. But that doesn't really talk about the flow of money for the network, which is sort of, how do Maine's compensate each other and so forth. And when I buy network service in my buying, you know,

is that the Sacco recursively until 4 to, the proposal was really agnostic on that. The the other proposal vanassche peering actually, one of the co-authors on that paper paper and that was purely pairwise. That was saying that rather than, you know, when two domains compare, the traffic rates today, do this solution. We're actually trying to put this into a larger context of the all of the players in the internet, who's all in for us. The Nash bargaining is really between a content provider and The Last Mile provider who is

trying to do termination fees and we're actually said the straight typical service contract between Directly, interacting pieces. That is filled last while providing the transit and content. I don't know whether that is address, is your question or not. If not rush you or ask you to follow up on slack. From Colgate asks, how does it compare to a nonprofit i-60 is the fundamental difference that is geographically distributed. Yes. Absolutely. Are. Clearly in the

paper that you can think of this is just so difficult XP and it has many of the same properties but I think by being Global is point is to transform. But absolutely it borrows the the concept of an ISP and tries to experiment with a pot or control over couldn't get their internet and how it Apocrypha interact with International boundary is given funding The pox are like you think of it as being bound to a particular and in fact that was one of the points of by having the pots terms of service, be where you so does enforce neutrality, you

are the less reliant on the local regulatory infrastructure for enforcing. Those kinds of things that doesn't you know we didn't talk about this issue just because if the ignition states that said you know you must sensor or whatever if you're going to have notes in our country how the POC response to that is an open question just like it is open for networks today. The last question, I'll cool off of the zoom g1x. The books are consumed in a pot, be implemented. What does that relationship look like to

introduce further advantages or challenges? In that case? So we we talked about cdn2 little bit. They're they're obviously crucially important and their two things that the pocket self could you do 40 DM to DF. Sedge providers can continue. The old requirement is that you were going to be the toughest edn's would have to be offered him on an open basis. Meaning that that there's a set price that and for that price, people can put their boxes in or they're open cdm's rather than allowing particular content of service providers have access and others not

I said Scott, thank you so much. Thank you everyone for listening. And there's some more following questions coming up his slack. And okay, thank you very much. I have a few questions. Do we have an example of a non-profit Global Network that's kind of stitch together? You know. Internet to Shay on Janet Serenity are all of the research and educational networks around the globe. How do you compare that with your your proposal met? Most of the nonprofit's at least in the US

connect to the educational institutions and so forth. And that's really not getting at. We want to make sure that last mile providers don't have any leverage over contest and service providers. And also, we want the global flow of money at least. So in the USBs are really connecting, not other. occasionally research institutions that other really not addressing fundamental economic problems where Network neutrality arises No. But at least they, they give us an idea of how a

network might be constructed in the sense that, you know, there is a, there's a one network in North America. There's networks and every European nation states and then their J on that loose them together another words, they ask that you actually have to construct these things. So I wasn't clear to me, you know, it seems like a pack would have the same kind of structure that you would have some sort of national structures and some wider, let's a european-wide structure on top of that. So, we were not thinking necessarily of having it along

National structures. I think we were presupposing. First of all, how to, to leverage should have how you might set up a an economic arrangement to Leverage large-scale. Least lines. I mean, you know, we've inside of the Australian Network, as an example of which we know that there are these nonprofit networks. We know we know how to manage them. We don't think that, you know, breaking new ground. I think it was the scale and it actually sort of having a

macroscopic impact on the overall economic organization of the internet. Okay. I mean I like I like the idea of the park. Why, why in your talk to you? So they say there's no bgp involve for example, because it seems like, you know what, you're going to build something today. You need a protocol to talk to other networks. So can you talk a little bit about that connected to it? The park, run some routing album probably wouldn't be your choice. There's no reason for policy while the educational

Network, San Jose and all that use bgp because you have to glue the thing together some way. And it's it's a way of of isolating administrative domains. It's the only one we have today. History of domains inside Transit Transit is not about administer that Claire that helps clarify, that have to reach agreement. Then you start getting this Cascade of a transitive hearing Arrangements. Transitive bgp and non transitive helps clarify. We have one question here from Reno. He says, any size or

specific property of Internet ecosystems that would make a more suitable for trying to talk. Nothing comes to mind. I think that the, you know, the hope is that we can start recently small. This will attend some of the questions on all this came up in the paper, the global. I x p. I mean, we're not inventing anything new. What we're doing here is analyzing its impact on network, but I don't know where this would first take. Hold of it. I think I would interpret this question of

if we were to build something like this, they're really two questions who's going to provide the bandwidth and then who were the first lnp's who would sign up for it. And that second question is a really good one and I don't know. The go-ahead. Yesterday, we had a question from Anya Feldman, which issue basically pointed out that there are these nonprofit ixp, you know, exchange Point all around the globe perhaps they could play some role in this development. That

the park Woods have terms of service is open. And as long as other networks, were willing to sort of abide by that they could attack him to wifey's would be a natural extension of this.

Cackle comments for the website

Buy this talk

Access to the talk “A Public Option for the Core”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “SIGCOMM 2020”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “IT & Technology”?

You might be interested in videos from this event

November 9 - 17, 2020
Online
50
94
future of ux, behavioral science, design engineering, design systems, design thinking process, new product, partnership, product design, the global experience summit 2020, ux research

Similar talks

Kyle Schomp
Performance Engineer, Senior II Lead at Akamai Technologies
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Dongzhu Xu
PhD Candidate at Beijing University of Posts and Telecommunications
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Tong LI
PhD Candidate at Huawei
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Buy this video

Video
Access to the talk “A Public Option for the Core”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
949 conferences
37757 speakers
14408 hours of content
Yotam Harchol
Dirk Bergemann
Nick Feamster
Eric Friedman