[00:00.000 --> 00:13.000]  Hi, everyone. This is I2P, major changes of the peer-to-peer network by Conrad Bechler.
[00:13.000 --> 00:17.000]  Close enough. Enjoy.
[00:17.000 --> 00:22.000]  Hello, everyone. Thank you for being here. That's my third I2P talk at FOSSTEM.
[00:22.000 --> 00:26.000]  First one was about DNS. That was the most difficult for me personally.
[00:26.000 --> 00:33.000]  And now a major changes of the peer-to-peer network of a bunch of programmers,
[00:33.000 --> 00:37.000]  which are creating I2P, which is an overlay network.
[00:37.000 --> 00:44.000]  I'm an independent researchers and software architect and developer,
[00:44.000 --> 00:52.000]  and I'm developing libraries, and I'm working full-time for Diva.Exchange,
[00:52.000 --> 00:56.000]  which is a small non-profit association in Switzerland.
[00:56.000 --> 01:00.000]  And there is nothing like a centralized model involved,
[01:00.000 --> 01:03.000]  because I'm totally living in the peer-to-peer I2P world,
[01:03.000 --> 01:08.000]  because I really love what the I2P developers and the open sources space are doing.
[01:08.000 --> 01:14.000]  Since the last 20 years, and I suggest we're looking now at the agenda.
[01:14.000 --> 01:21.000]  I'll give you a short introduction of I2P, and then we're looking at what the developers did.
[01:21.000 --> 01:25.000]  As the latest changes of the code.
[01:25.000 --> 01:30.000]  And we'll focus, since I'm an application developer based on the I2P network,
[01:30.000 --> 01:33.000]  what is the impact? We're looking at the summary.
[01:33.000 --> 01:38.000]  And I understood that there are no questions, so if you have questions,
[01:38.000 --> 01:41.000]  I'll be outside in the hall for a few minutes,
[01:41.000 --> 01:49.000]  and I'm happy to get in touch with you if you have critical and skeptical questions.
[01:49.000 --> 01:53.000]  Short thing, what is the role of Diva.Exchange within I2P?
[01:53.000 --> 01:55.000]  I2P is a developer team.
[01:55.000 --> 01:57.000]  Diva.Exchange is a developer team.
[01:57.000 --> 02:01.000]  And the I2P developers are spread all over the world.
[02:01.000 --> 02:04.000]  The Diva.Developers are spread all over the world.
[02:04.000 --> 02:08.000]  We're cooperating friendly, but we're not the same.
[02:08.000 --> 02:13.000]  Diva.Exchange is primarily a library developer and is supporting
[02:13.000 --> 02:19.000]  official, so-called official, open-source Docker images.
[02:19.000 --> 02:20.000]  That's what we're doing.
[02:20.000 --> 02:24.000]  Myself, since I'm a researcher, I'm cooperating with Swiss universities,
[02:24.000 --> 02:31.000]  and I'm open for all corporations with any university to take a very skeptical perspective
[02:31.000 --> 02:34.000]  towards our own work.
[02:34.000 --> 02:36.000]  Now what is the I2P network?
[02:36.000 --> 02:39.000]  I2P network is an overlay network.
[02:39.000 --> 02:44.000]  And now please jump to the very last line of this slide.
[02:44.000 --> 02:51.000]  I2P gives you privacy by design, serious privacy by design.
[02:51.000 --> 02:55.000]  It means you get confidentiality and total anonymity.
[02:55.000 --> 03:02.000]  Just as an example, this statement that the network gives you total anonymity
[03:02.000 --> 03:08.000]  is something which we do research in Switzerland, whether we can break this anonymity.
[03:08.000 --> 03:10.000]  And that's one of our research approaches.
[03:10.000 --> 03:16.000]  But currently, at the current state of knowledge, and since I2P is existing
[03:16.000 --> 03:23.000]  since 20 years, we can say it's confidential and anonymous.
[03:23.000 --> 03:24.000]  Now a question.
[03:24.000 --> 03:29.000]  Whoever in this room, in the past presentations, there was around 5% of the audience
[03:29.000 --> 03:33.000]  have ever gotten in touch and used I2P in their life.
[03:33.000 --> 03:38.000]  Please, hands up. Wow, that's close to 20%.
[03:38.000 --> 03:40.000]  Cool, thank you.
[03:40.000 --> 03:46.000]  So for the others which do not know it, anonymity has a cost attached.
[03:46.000 --> 03:49.000]  And this cost is called latency.
[03:49.000 --> 03:54.000]  So when I'm talking about an anonymous network which gives you full confidentiality,
[03:54.000 --> 04:02.000]  then these messages are multiple time encrypted and hop from peer to peer to its destination.
[04:02.000 --> 04:07.000]  And through another tunnel, they hop back, the reply hops back.
[04:07.000 --> 04:14.000]  And this is giving anonymity on a very, very, very high level.
[04:14.000 --> 04:16.000]  A word to be clear here.
[04:16.000 --> 04:21.000]  I2P has no storage capability by itself.
[04:21.000 --> 04:25.000]  It's a transport layer.
[04:25.000 --> 04:28.000]  Now how to get I2P?
[04:28.000 --> 04:31.000]  Those which got in touch with it, they know that.
[04:31.000 --> 04:34.000]  There are the Linux repos, and then there are PPAs,
[04:34.000 --> 04:39.000]  and then there are the container maintained by Devo Exchange, which you find on Docker.
[04:39.000 --> 04:41.000]  Now why is there I2P?
[04:41.000 --> 04:43.000]  And why is there I2PD?
[04:43.000 --> 04:49.000]  Because there are two versions available from two development teams.
[04:49.000 --> 04:51.000]  In short, I2P is written in Java.
[04:51.000 --> 04:53.000]  It has a user interface.
[04:53.000 --> 04:58.000]  And there are a lot of fans around there which really like that.
[04:58.000 --> 05:03.000]  And then there is a smaller version, which is I2PD, which is written in C++,
[05:03.000 --> 05:07.000]  and which is the demon only, so it's rather a bit more for admins.
[05:07.000 --> 05:13.000]  Devo uses I2PD, and there's also the image maintainer on Docker for this version.
[05:13.000 --> 05:17.000]  From my point of view, both are equally valid.
[05:17.000 --> 05:22.000]  Word of warning, please don't trust any binaries which floating around here and there,
[05:22.000 --> 05:29.000]  which you can't reproduce locally by yourself, please build it yourself if you can, and sure you can.
[05:29.000 --> 05:34.000]  Because this gives power to the open-source developers.
[05:34.000 --> 05:38.000]  So if you just consume some binaries, please don't.
[05:38.000 --> 05:44.000]  It's also dangerous because you have a router running on your local machine.
[05:44.000 --> 05:49.000]  The whole thing here is it's a peer-to-peer network without any central authority,
[05:49.000 --> 05:53.000]  without any trust involved.
[05:53.000 --> 05:58.000]  Now, the latest changes.
[05:58.000 --> 06:03.000]  Since 20 years I2P is existing and gets developed,
[06:03.000 --> 06:11.000]  and there are two transport protocols.
[06:11.000 --> 06:14.000]  TCP and UDP.
[06:14.000 --> 06:20.000]  TCP as also used within other overlay networks was a long while,
[06:20.000 --> 06:28.000]  so around four years already upgraded to so-called NTCP2, which is much modernized,
[06:28.000 --> 06:33.000]  so the TCP communication was already pretty well done in the last four years.
[06:33.000 --> 06:39.000]  But UDP, which is something where you just blow out messages to the network,
[06:39.000 --> 06:46.000]  and I'll talk about that in a minute, which is fast for I2P in an I2P context,
[06:46.000 --> 06:50.000]  was not modernized since 15 years.
[06:50.000 --> 06:56.000]  And the developer of I2P, they worked one full year on modernizing UDP,
[06:56.000 --> 06:59.000]  and they call it SSU2.
[06:59.000 --> 07:06.000]  So when you hear SSU2 in the context of I2P, it's UDP.
[07:06.000 --> 07:13.000]  If you want to dig deeper, then please take a look at noiseprotocol.org,
[07:13.000 --> 07:19.000]  that's very important also for other overlay networks you might have heard of,
[07:19.000 --> 07:27.000]  and the developer team of I2P has borrowed quite some things from WireGuard,
[07:27.000 --> 07:32.000]  VPN, and also from Qwik, and there are some RFCs around,
[07:32.000 --> 07:37.000]  so please dig deeper by using these hints.
[07:37.000 --> 07:42.000]  Then the cryptography had to be updated too.
[07:42.000 --> 07:47.000]  There will be more work coming in the upcoming years, I know for sure.
[07:47.000 --> 07:51.000]  But it's now already on a quite good state.
[07:51.000 --> 07:53.000]  All right, seven minutes to go.
[07:53.000 --> 07:56.000]  What were the goals?
[07:56.000 --> 08:01.000]  UDP, the performance, is simply much better in an overlay network
[08:01.000 --> 08:05.000]  than TCP.
[08:05.000 --> 08:12.000]  Me as an application developer, which is creating a fully distributed storage layer
[08:12.000 --> 08:15.000]  on top of the I2P network,
[08:15.000 --> 08:22.000]  performance to communicate between the nodes in the I2P network is an issue.
[08:22.000 --> 08:27.000]  So for me, it was always the choice to take UDP,
[08:27.000 --> 08:35.000]  and I implemented a gossiping protocol to realize this fully anonymous blockchain,
[08:35.000 --> 08:38.000]  and so this distributed storage layer,
[08:38.000 --> 08:44.000]  and I needed UDP that the peers could communicate within each other.
[08:44.000 --> 08:51.000]  So for me, this was really an important goal that they could improve the performance a lot.
[08:51.000 --> 08:57.000]  Nationally, as you know, there are countries in this world where communication is not that easy
[08:57.000 --> 09:00.000]  as we have the luck to be here in Europe,
[09:00.000 --> 09:03.000]  so obfuscation is a topic.
[09:03.000 --> 09:07.000]  For all those people, for all the whistleblowers and the journalists
[09:07.000 --> 09:11.000]  which are using I2P in critical countries,
[09:11.000 --> 09:13.000]  they must not be detected,
[09:13.000 --> 09:16.000]  and I2P is one of the premier solutions too
[09:16.000 --> 09:23.000]  for such people which need communication on a very confidential level.
[09:23.000 --> 09:25.000]  So the obfuscation was a big topic,
[09:25.000 --> 09:33.000]  and additionally, again, UDP is easily attackable,
[09:33.000 --> 09:37.000]  so the developers had to look into the security problems they had,
[09:37.000 --> 09:41.000]  and they did, and they did a good job.
[09:41.000 --> 09:46.000]  Right.
[09:46.000 --> 09:50.000]  There are a few other challenges of UDP, like the fragmentation of messages.
[09:50.000 --> 09:54.000]  You know, you have a long message, you want to send it through the network,
[09:54.000 --> 09:58.000]  but it gets chopped up into pieces, and UDP is not reliable,
[09:58.000 --> 10:04.000]  so the developers really had some challenges to solve in the last 12 months.
[10:04.000 --> 10:09.000]  But the fact is, they have implemented strong solutions,
[10:09.000 --> 10:13.000]  and today, if you look at the release notes of I2P,
[10:13.000 --> 10:20.000]  I2P is in a much better state.
[10:20.000 --> 10:29.000]  And this is what I want you as a core take out of this presentation.
[10:29.000 --> 10:35.000]  I2P gives you, since 20 years, real privacy by design,
[10:35.000 --> 10:38.000]  and the developers did a good job the last 20 years.
[10:38.000 --> 10:42.000]  But now, with this performance improvement,
[10:42.000 --> 10:46.000]  and that's my personal view on their work,
[10:46.000 --> 10:49.000]  they really made a major step forward.
[10:49.000 --> 10:56.000]  Because UDP is for application developers really important within the I2P network.
[10:56.000 --> 10:58.000]  That's my personal view on it.
[10:58.000 --> 11:02.000]  Please feel free to criticize me afterwards in the Q&A session outside
[11:02.000 --> 11:06.000]  if you see this differently, or if you have a different view on it.
[11:06.000 --> 11:08.000]  I'm happy to hear that.
[11:08.000 --> 11:16.000]  But these reductions, they are relevant.
[11:16.000 --> 11:23.000]  For us as application developers, we directly felt this performance impact.
[11:23.000 --> 11:27.000]  In November, when the release was, so four months ago,
[11:27.000 --> 11:29.000]  or three and a half months ago,
[11:29.000 --> 11:35.000]  the performance in our blockchain test network improved very much.
[11:35.000 --> 11:39.000]  And although I believe from research point of view,
[11:39.000 --> 11:44.000]  we have quite some work ahead of us to,
[11:44.000 --> 11:49.000]  skeptically and very critically looking at the latest changes
[11:49.000 --> 11:53.000]  to see whether some weird box have been introduced.
[11:53.000 --> 11:56.000]  The first impression of the two months is,
[11:56.000 --> 11:59.000]  that's a new generation of I2P.
[11:59.000 --> 12:04.000]  So for me, the performance impact is really, really relevant.
[12:04.000 --> 12:08.000]  We're on the target.
[12:08.000 --> 12:10.000]  We're seeing the target line.
[12:10.000 --> 12:15.000]  Takeouts.
[12:15.000 --> 12:23.000]  After 20 years of development, I2P is the leading peer-to-peer privacy by design network
[12:23.000 --> 12:25.000]  which is a transport layer only,
[12:25.000 --> 12:28.000]  which gives you today very modern cryptography,
[12:28.000 --> 12:36.000]  very modern possibilities to use as an application developer.
[12:36.000 --> 12:48.000]  The performance increased significantly since last November.
[12:48.000 --> 12:52.000]  And if you feel as a researcher too,
[12:52.000 --> 12:57.000]  or as an application developer to take a closer look at I2P,
[12:57.000 --> 12:59.000]  now it's a good time.
[12:59.000 --> 13:04.000]  Either you can invest as a researcher trying to break the anonymity
[13:04.000 --> 13:08.000]  and present it at an upcoming conference,
[13:08.000 --> 13:11.000]  or you try to create some applications on top of it.
[13:11.000 --> 13:16.000]  I believe currently is a good point to take a look again at I2P
[13:16.000 --> 13:22.000]  because now it's really quite fast to use and it's also fun to use.
[13:22.000 --> 13:27.000]  On GitHub, so that's in the lower right corner and you'll find it online on GitHub,
[13:27.000 --> 13:34.000]  you'll find Docker images which help you to get started with I2P really quickly.
[13:34.000 --> 13:36.000]  You can, because a peer-to-peer network,
[13:36.000 --> 13:39.000]  to create it as a test network is quite complex.
[13:39.000 --> 13:44.000]  But we created here simple images for you which you can start with Docker Compose.
[13:44.000 --> 13:46.000]  You have all the containers running,
[13:46.000 --> 13:57.000]  your member of the I2P network and you can write your application on top of this network.
[13:57.000 --> 13:59.000]  Here is a list of sources.
[13:59.000 --> 14:03.000]  Please do your own research too
[14:03.000 --> 14:12.000]  because privacy by design also means that we hope to motivate US developers
[14:12.000 --> 14:17.000]  to take a look, a very close look at the documentation,
[14:17.000 --> 14:20.000]  a very close look at the source code to be skeptical
[14:20.000 --> 14:27.000]  because privacy is such an important topic for us as developers.
[14:27.000 --> 14:32.000]  So we're always very, very pleased if we get critical feedbacks
[14:32.000 --> 14:36.000]  on the development of the stuff we're doing.
[14:36.000 --> 14:39.000]  Thank you very much for your time.
[14:39.000 --> 14:42.000]  Have a great 4STEM. Thank you.
[14:48.000 --> 14:51.000]  Thanks, Koran. Really nice talk.
[14:51.000 --> 14:56.000]  Again, if you have questions, please grab them outside and be happy to answer.
[14:56.000 --> 14:58.000]  I hope.
[14:58.000 --> 15:10.000]  Thank you.