[MUSIC]
 [MUSIC]
 Hello everybody, welcome to the last talk in this block.
 Who of you sent your stuff to the camp with BGP?
 Any user of BGP?
 I used it, it works great.
 This is the routing protocol at camp for parcels.
 But there's another BGP,
 which is the routing protocol on the Internet for packets.
 It's what has enabled the Internet for decades.
 Mentor is going to tell us about what might replace BGP on the Internet,
 and how Scion might actually help to make the Internet greener.
 Mentor, stage is all yours.
 Thank you.
 [APPLAUSE]
 In the next 20 minutes or so,
 I will talk about how the Internet actually works currently,
 and what the Scion project in particular wants to change about it,
 and how that works.
 I will gloss over a lot of the technical detail, unfortunately,
 because I have very little time.
 And I will try to be inclusive in this talk,
 which is why I will use metaphors that might stretch reality a bit.
 I simply want to give you, those of you who have no idea actually
 how the Internet works, some idea of how it does.
 And for this reason, please bear with me if you have an idea
 what I'm talking about, and the analogy doesn't quite fit.
 All right.
 So this is the outline of my talk.
 First, we need to talk about how the Internet currently works.
 The status quo is that the Internet pretty much works
 like a run-of-the-mill postal system.
 Say, I want to send this plush owl to the Chaos Computer Club in Berlin.
 So what do I do?
 I simply put it in a box and add a sticker to that box
 with a destination where I write the postal address
 of the Chaos Computer Club Berlin,
 give it to the nice folks at the BGP,
 I think they're called the Camp Parcel Service or something,
 and then they worry about the rest.
 I don't actually know the route which this packet will take.
 I only know that at some point in the future,
 I think a few days or so, if everything works as expected,
 my owl will arrive safely at Chaos Computer Club Berlin.
 So I don't know whether this packet will be forwarded
 via ZNNIC or via Fürstenberg.
 Maybe it can also take a detour through a logistics center
 in Frankfurt Oder.
 I simply don't know.
 Interesting.
 I think I need to restart my slides
 because I just rebuilt them.
 A short timer, please.
 Here we go.
 So this is the perspective of the user that sends a packet
 over the current Internet.
 They simply know where to put the packet, the next postal office,
 and some magic happens in between
 and the packet arrives at its destination at some point.
 How does the postal system do this?
 Well, in very simple terms,
 and this is unfortunately not exactly how it works,
 unfortunately, let's assume that there's a post office
 in every town and each has a few courier connections
 to the surrounding post office in neighboring towns.
 And each post office also keeps a huge list of all the towns
 in the country where it can look up into which neighboring post office
 it should send a packet in order for the packet to move closer
 to its destination.
 So we send a parcel, this parcel will get forwarded intuitively
 step by step close to its destination
 until at some point it arrives in its destination city
 where the postman knows exactly how to find the street address
 for this packet.
 So this distributed mechanism, this is pretty similar
 to how postal systems used to work in the past for several hundreds of years
 and works very well.
 A given post office only needs to know in which direction
 it needs to forward this packet and at some point the packet
 will arrive actually in the city where it is intended to arrive in.
 So in there it will be given to a postman or woman or person
 who will know how to deliver this packet.
 So resolving this analogy, the post offices are the internet core routers,
 the post offices.
 The local post office is your ISP which is where you will send your letters
 or give up your letters and receive new letters.
 And all post offices along the way are their own networks,
 their own autonomous systems, you have probably heard this term already.
 And all the neighboring post offices of a given post office are called its peers.
 And this huge list that each post office needs to maintain
 in order to know along which next hop a given packet needs to be sent
 is a routing table.
 And the BGP protocol, not those guys over there, but the BGP protocol
 is used to keep the routing table up to date.
 So what is wrong with this current internet?
 Well, depending on your perspective, not that much or maybe quite a lot.
 So the internet doesn't have the best reliability.
 So it has like three nines.
 It has a 99.9% reliability which sounds a lot but actually isn't that much.
 The good news about the internet, however, is that failures often last a few seconds
 which all of you have probably noticed that the website doesn't load,
 you hit F5 and it does load.
 And you don't know where the problem was, but it's pretty likely that the problem
 was with this BGP mechanism where a failure wasn't needed to be resolved
 in order for routing to work again.
 There are other problems like security problems because BGP as a protocol
 was designed at a time where there was an implicit assumption
 that every participant on the network can be trusted.
 Actually in the BGP RFCs, in the first ones, the words trust or security
 or malicious or attack or something, you can grab for this and zero hits every time.
 The BGP was designed with a completely different internet in mind than we have today.
 Then there are some broken economic incentives.
 So as we all know, currently legislation, politics is lagging behind
 and they don't really enforce a proper carbon tax on carbon emissions
 which means that it's still the case today that you can buy electricity
 from non-renewable energy at lower prices than electricity
 that was created using renewable energy.
 And obviously market forces force network operators which have running costs
 and need this energy, force them to basically just buy the cheapest one,
 buy the discount of electricity.
 And this is obviously a problem, this is one of the points where the internet
 is currently not as green as it could be.
 But it's also hard for users to influence, like you don't really have a choice
 when some network in between decides to buy electricity from coal plants.
 You're basically stuck with this.
 Now, what is different with Scion?
 A lot of things they are trying to address, a lot of these problems,
 for instance also addressing security aspects, but I won't go into detail here.
 The key point where Scion differs from the BGP-based internet is that
 with Scion the senders get to decide along which paths the data gets forwarded.
 So again in our postal analogy, what we do instead of writing
 a regular street address there, we write steps there.
 We say, okay, the first step is the camp parcel service, the packet needs to go there,
 before the camp parcel service forwards it to Mildendorf,
 before Mildendorf forwards it to St. Annick and so on.
 And at some point it arrives at a point where a post person, a postman or post woman
 can simply read the street address and exactly knows where it is.
 So this is the postal analogy.
 How does this help?
 How, translating this to data packets, what do we gain?
 We gain as users the ability to boycott networks that we know operate
 with a dubious sustainability record.
 And we can also support networks that we know operate sustainably,
 that have basically given assurances to other parties
 that they are only buying energy from renewable sources.
 And in our example, when we want to send this owl to Berlin,
 and we know that the operator of the connection between Ziedernick and Oranienburg,
 which is on the way to Berlin, uses a truck,
 and we also know that the operator that sends traffic or packets or parcels
 between Fürstenbeck and Oranienburg uses rail,
 then I can, as a user decide, please route my packets along Fürstenbeck,
 even if it takes a bit longer.
 But I know that the carbon footprint of this transaction will be smaller.
 There's an additional feature, of course, that I should mention.
 This allows all sorts of things to do with this capability.
 For instance, for activists it's also interesting maybe to avoid
 routing via networks that are under dubious jurisdictions, etc.
 So you can optimize for more things.
 It's also interesting that CYON has been designed in a way
 that allows its core routers to operate more efficiently.
 In the post office analogy, it doesn't really keep these long lists
 with which it needs to decide where to send on which packet,
 it doesn't need to keep this list of all the towns in the country.
 It simply needs to follow the instructions that are written on the parcel.
 This comes at the expense, however, of 5 to 20% of bandwidth overhead,
 which sounds like a lot.
 It's not strictly an issue in wired communication,
 because increasing the speed doesn't really scale the electricity need of this.
 If you just put a new laser with a different wavelength on the same fiber,
 the laser doesn't really need more energy in order to transport more data.
 But to basically visualize this overhead that this introduces,
 imagine putting this owl into a parcel where you write,
 I don't know, the BGP guys over there onto it,
 and then put it into another parcel and write the next address.
 Actually, it's the other way around.
 First, you would write the address of Berlin,
 put it into the next parcel, which you close,
 and then you write Oranienburg there, and then put it in a parcel,
 and then you write, I don't know, what did I say?
 Christenberg on it, put it in a parcel,
 and then on the last parcel you write BGP,
 and then they unwrap it, and so on.
 This obviously includes some overhead,
 and this is a visualization of what kind of overhead it entails.
 The technical details are a bit different, but this is how it works.
 Also, there's something that we should mention,
 is there's an embodied CO2 footprint,
 because SIRON router hardware is different from hardware
 that needs to run the regular internet.
 And obviously, rolling out SIRON globally would mean
 investing this, embodying this CO2 footprint of the data
 and constructing all this hardware.
 Actually, numbers, how bad this will be, I'm missing,
 I don't know about this.
 I think a rough ballpark is that a SIRON router will be about,
 somebody about as much energy as an equivalent IP internet router.
 So maybe this sort of evens out,
 but you could feel free to research this yourself.
 A third point which is actively being discussed in the SIRON project
 is to attach carbon footprint of each individual hop
 as metadata to the routing information
 that then allows the user to construct paths
 that use the minimal amount of CO2,
 and then let the user preferences cause pressure on network operators
 to use renewable energy or to prefer the use of renewable energy.
 Another aspect of this would be that if they are being honest about
 their current CO2 emissions used for routing their packets,
 and this changes dynamically basically,
 for example based on available wind,
 then using this information could cause,
 could help us route our packets around,
 I don't know, how is it called, if the wind is gone,
 around calm weather patches where there is little wind,
 too little wind power for the wind turbines,
 and so we simply avoid these regions and send our packets by different routes.
 All in all this could sort of create a virtual circle
 where the network operators try to compete with each other
 and try to be more green than their competitors,
 basically reducing carbon emissions overall.
 Will this be enough? Will this really help us fighting climate change?
 Well, so the global internet energy consumption in 2020 was 207 terawatt hours,
 which is sort of creates to a momentary power consumption of about 31 gigawatts,
 which is double that of the biggest hydroelectric plant in the world.
 The Three Gorges Dam in China produces about half this amount of power
 if it's running at full capacity,
 and this is excluding the power consumption of data centers,
 so this is only network equipment, it's a rough ballpark figure.
 And this number is expected to grow to about,
 so right now it's 1% of all the global electricity,
 and this number is expected to grow to 10% by 2030.
 So the bottom line is we don't really know how much Scion will help.
 We know that taking all the efficiency gains of routers into account
 and subtracting the bandwidth overhead that it has,
 we end up at a number of around 10% more efficient,
 10% less electricity use, let's say.
 Of course, we didn't really take into account the embodied carbon costs,
 and we also didn't really take into account the currently still elevated overhead,
 energy overhead that is required for anti-virus to actually talk this protocol.
 I mean, you can help.
 So it probably wouldn't be enough to save the planet by itself, right?
 But it's one step along the way,
 and we can't really afford to exclude things that don't promise that much reward,
 but we can probably use this as an argument to still push towards a technology
 that changes the way that the Internet is currently used
 and at the same time empowers users in a way that allows them,
 allows additional forces to push towards carbon neutrality.
 In this talk, I have left out a bunch of stuff which we could talk about offline afterwards.
 Of course, there's a lot of technical detail that I glossed over pretty much.
 There are questions about how a SCI-internet would be governed.
 There are also different incentives for different Internet stakeholders.
 Your ISP might not be that happy if you always decide to send your traffic a certain way,
 because right now it's them who decide how to forward it.
 Anyway, how to get involved?
 So the SCI-in project itself has been under development at ETH Zurich since 2009.
 It's open source and standardization is underway at ITF also.
 It's deployed in the wild.
 The Swiss Secure Finance Network will migrate exclusively to SCI-ins,
 so all the Swiss banks will stop using the old system and exclusively migrate to this system by 2024.
 There are additional networks.
 So one is a production network that is run by a commercial company that is selling this hardware also.
 And an education and research network that you guys can join.
 And also there's an experimental playground where you can play around with this technology,
 but it still uses the regular BGP-based Internet underneath and works via VPN links.
 You can play around with this under these URLs and get involved.
 It's probably best if you join this Matrix channel where SCI-in development is discussed.
 And I think with that I'm about done.
 You can also reach out to me at this email or this masternode address.
 [Applause]
 [Music]