New DEMO: SWISS SUBNET pushing Mass Adoption of ICP Cloud Engines

本站添加: 2026-06-03

[00:00:00]And we are working closely with Dfinity to shape what is needed to take cloud engines from the small pilot that is focused on to something that's really production ready for corporates and governments around the globe.

[00:00:09]>> Hello guys, welcome to zero to hero. As you have noticed from the teaser of this video, today we're going to discuss about the Swiss Subnet and I will show you a little demo regarding the technical improvements regarding the integration of cloud engines, the client pipeline, and how, as mentioned, they're planning to onboard a lot of large institutions and governments that are going to use these services moving forward. Very soon, I will have a meaningful announcement regarding the Swiss Subnet. I will have more news.

[00:00:41]Stay tuned for that. And now, let's check this video together.

[00:00:45]>> Hi, my name is Markus and I'm the CTO of Swiss Subnet.

[00:00:48]Let me see if I can take over sharing.

[00:00:51]Yes, looks good. So, back in 2025, we were the first company ever to rent an ICP subnet.

[00:00:58]And in the meantime, we're up and running in a closed beta phase. Our clients are corporates and government entities who want to build on a self-sovereign cloud stack, but are afraid to do this in a wild west of public and permissionless blockchains.

[00:01:09]So, our clients want the benefits of the ICP technology, but they also need help starting their journey onto this brand new blockchain-based technology stack.

[00:01:17]And we help them to walk before they can run.

[00:01:21]As part of setting up the Swiss Subnet, we work with 13 independent professional data center and IT service companies in Switzerland and Liechtenstein to become the node providers. And you see this when you go to the dashboard of the Swiss Subnet, uh all spread across Switzerland and Liechtenstein. You can see more about the node providers by clicking on this.

[00:01:40]Uh these are mostly data center companies uh who want to be part of the future of computing, so ICP and blockchain.

[00:01:46]And also, uh some are professional uh IT service companies in Switzerland and Liechtenstein who are looking to onboard their clients uh to the Swiss Subnet.

[00:01:55]And uh I can actually share data analysis experience from Pakistan, also in Switzerland and Liechtenstein. IPv6 is the biggest hurdle in deploying ICP to small to medium data centers.

[00:02:07]All those we subnet providers have Gen 2 SCV nodes.

[00:02:10]We are the only subnet that has Gen 2 nodes exclusively.

[00:02:14]And this enables us to become the first fully encrypted SCV subnet.

[00:02:19]That's because confidentiality and integrity are obviously very important to our clients.

[00:02:25]Since 20 26 January this year, we've been in closed beta with more than 100 developers.

[00:02:32]And I want to tell a little bit about how we're targeting our clients uh before I hand over to Quint for a demo.

[00:02:38]So, there's not going to be any slides because our entire focus in Swiss subnet really is on the developer experience for small to medium corporates and governments. And for them, the journey doesn't stop when the canister is created. The having a canister online is just the start.

[00:02:50]The developers uh for these corporates, they need to know what their canister is doing, uh how they are are growing, which ones are misbehaving. Uh they also need to be able to shut down their own canisters because in our private subnets, there are no cycles that they can run out of. So, if they have an infinite loop and just keep processing, they can actually starve uh all other canisters. And they will never run out of cycles, so they will keep starving until someone shuts them down.

[00:03:14]So, that instrumentation is needed and that's what we're building into the developer console and the admin console for Swiss subnet.

[00:03:21]Quint will walk you through this in a minute.

[00:03:24]We expect that our bigger clients will have their own dedicated SRE teams, so we provide all the tooling that they need to keep their rental subnet infrastructure up and running.

[00:03:33]Two more things that might be interesting for you is that we will also support multi-tenant subnets so that we can uh shred the capacity of a subnet in pieces that we rent out to different clients.

[00:03:44]And then we obviously want them to be able to work in isolation, so we'll have tooling in place to manage each other's resources and to give them control over their canisters.

[00:03:51]As um corporates and governments uh and sometimes having multiple developer teams, they expect comfort features like being able to group their canisters, being able to have different development teams with different access rights. We provide all of this in our admin console and our developer console.

[00:04:07]As I said, we started last year with the subnet rental canister uh framework, but today obviously we have cloud engines available as well and we're excited to the outlook that this gives us as we will no longer be limited to the one-size-fits-all approach of a 13-node physical subnet, but we can offer our clients a smaller subnet. We can offer our clients to grow from a small subnet to a bigger or to resize from big to small, whichever way they go.

[00:04:30]We also have the perspective to hand over control of a cloud engine to our client once they're ready to start running instead of just walking with our help.

[00:04:38]And we're working closely with Dfinity to shape what is needed to take cloud engines from the small pilots that is focused on to something that's really production-ready for corporates and governments around the globe.

[00:04:48]Let me stop talking here and hand it over to Quint.

[00:04:50]>> I will go um zoom out a little bit again before I start. So one of the pillars we've been focusing as on on Swiss Swiss subnet is basically managing our own rental subnet, right? So for this we basically started with building in-house tooling to make sure we can get a view of what's running, who runs what, who gets access to what, make sure our users have all the tools they need. So I'm going to try to speed run a little and I'm going to do it live. I don't have a prerecorded video sadly. It's a good idea. I'm going to do this next time. Um but so basically what I'm going to do is I'm going to show how we manage the Swiss subnet ourselves. I'm going to show this do do it through our staging environments. So if the numbers look look a little low um and this is because it it's under our development environment and even today um we upgraded it already five times so I hope I didn't break anything uh by doing this. Um I'm going to section our approaching in in three big parts. Um the the first part is So, as we subnet, we are uh um Yeah, we have a rental subnet. So, as a rental subnet owner, we have we we need to make sure to maintain what's on there. We saw by recent conversations around Cloud Engine that this concept maps very well with how Cloud Engines will be run. The Cloud Engines will be either a single tenant, so you rent the whole Cloud Engine, or if you rent the Cloud Engine, you can also, of course, onboard different parties, different entities on there, and then you're going to be in multi-tenant, which is basically what we're currently running under.

[00:06:20]Uh and so, for this, we need three big pieces. One is, of course, a tool for these um administrators of the Cloud Engine, of the the subnet rental. And then, one level lower, you have either your development teams, development teams of others. Um, so basically, for this, we took inspiration from Orbit, which uses asset management, but can also basically do canister management. So, we have those concepts, but we go a little little broader and trying to make it easy. And finally, of course, we have the developer experience. So, we want to make it as easy as possible to set things up and get running. So, I'm going to try to just download the zip from our front end, do ICPD deploy with the new CLI, and and let's hope it magically works. Um Aside from that, we're also working on some POCs, for example, Terraform Ansible integration to actually maintain and, for example, define a canister as a a Terraform resource uh in your infrastructure as code, um which uh yeah, most enterprises are used to uh using. So, first thing's first, uh so, as a user, uh at least that's our approach on onboarding uh while we're in beta is we basically go to the console. This is staging. You won't You won't be able to log in here, but if you just go to subnet.ch, you'll find your way around if you want to join the the the public beta we're doing, and we, of course, can sign up from the waiting list. So, we have currently a waiting list, which will be removed in the future. And here, I am going to Thanks for all these AI features, enable multiple accounts to make sure that I have a fresh one cuz I tried this out before, of course. So, this is going to give me a a unique delegate in the front end.

[00:07:53]I am going to log in and I can see, "Hey, you're on the waiting list.

[00:07:58]Uh can you give your email?" I'm going to for the demo skip email verification, but it says, "We'll be in touch." So, how this looks from our side is basically we have an admin. I hope you can also see um this page. This is again this is staging. These are not real numbers.

[00:08:11]It's mostly 15 or 18 times me uh and other engineers on our teams who have been testing this out. But, if I go quickly go back and of course Zoom is always in the way. I can basically get my user ID to find where I am in this list.

[00:08:25]And I can do a quick search. Of course, I have to uh make sure that I can find myself. I can activate myself. And just to see that it works, I can refresh this page and it should say, "Hey, you're in." Perfect. And let me run through like what you can see here.

[00:08:39]So, this is the back office, the back of the back office. So, it's not about organizing teams and developing canisters. This is really about having insight into what's on the on your subnet, on your cloud engine, and what happens there. Um we have notions for integration partners. We call them here trusted partners. So, for example, uh Marcus did some pilots with Caffeine AI uh a few months back that Caffeine basically could directly deploy um to the Swiss subnets. So, this is just a notion of this integration. And of course, as an admin, you want to fine-tune some things. So, we for internal management, we have users and we have some roles reading organization, billing. And what I mean is of course I can see all the test organizations we already created uh during testing.

[00:09:18]So, this is one part. And of course, you have a list of the users like I showed before. And you can basically gate the initial onboarding. Now, the moment users are in, you need to basically see this as just a sign-up. You log in to GitHub. And the moment you're in, you can start creating organization. And this is where we go to the second level canister uh management. So, to make this easy, you get a default organization. We can go to settings, of course. In our case, we have a terms and conditions.

[00:09:42]Let's quickly accept this. We have no canisters yet, but in settings So, we have a free tier. We have different payment tiers, and we also integrate payment providers directly. We have a notion of teams and projects. So, teams are how you manage users, and projects is how you group canisters.

[00:09:58]Um so, you can make easily just apply certain rules, certain permissions on a group of canister all at once.

[00:10:04]I will not go into detail. Um if you ever want an in-depth demo, you can always go to the official website and uh get into contact with us. We also have an invitation flow. So, either you can invite someone by email, the user ID I used before, or the principal they used to to log in with. It doesn't really matter. Just the nuance of what identifier you prefer for the user.

[00:10:22]And then, of course, we have a project, we have canisters, and in project settings, you have teams. And teams you can also fine-tune. So, on this level, you can also see, for example, the default team, which is me. I'm the only user. I can do everything I want, but I can create a team that can only read the canisters, so they cannot start and stop anything.

[00:10:38]Aside from that, maybe if you're familiar with Orbit, you can do way more than just pressing a button and something happens. We have basically built in approval policies and audit logs. So, this would mean if I configure if you want to change or add a controller to a canister, to create a canister, you need two two votes at a minimum. Two out of three. You need this percentage of the team that basically signs off, yes, you can create this. And this can be extended to other canister operations, and you can all configure this like straight from the UI. We have a little dashboard, but I'll I'll show some more interesting graphics.

[00:11:09]Um and so, basically, we can go to canisters. Um this is all live, so our staging lives on the Swiss subnets. Um so, I'll just create one.

[00:11:18]Uh this is created. I'll call it canister A. Um Let's save this. And then, we of course, can create another canister.

[00:11:27]Um Let's call it canister B.

[00:11:31]You can also rename your organizations, of course, to make it more readable. And this is where I think we looked at the new CLI and we're a fan of how easy it is to use. So, we basically made use kind of the same notion as bundles, but the other way around instead of pre-bundling, we're just bundling the whole projects and pre-configuring things for you. So, this is where my demo is probably going to go wrong, but I hope it's not. So, what you can do is just download the project project and it's going to download the zip. You can see I downloaded the zip before, so it's called default project, the same as you. And it will basically encapsulate the old canisters. Let me go to my second screen, quickly unzip this, go into the project, and open a visual editor for everyone. I hope you can also see VS code.

[00:12:10]So, for now, it's a very simple read me, but the nice thing it's all pre-configured. So, if you do icp.yaml, we use the recipes that the foundation provides us, but it also pre-configures all your all your IDs. So, it's a direct reflection to easily get started from on the console straight to your CLI of development environment.

[00:12:28]Now, I hope I have it installed. So, it should be as easy as icp deploy. Raymond also touched upon, so there's different environments.

[00:12:36]The nice thing here, for example, we use it internally to define our staging and production environments.

[00:12:42]Since they're all both mainnets, but we configure for configured them differently and there's other assumptions when they're pushed. So, if we just choose to dfx deploy, of course, it says I am not um a controller, which makes sense. So, if I just quickly get my principal here and go back to our console.

[00:13:00]Of course, I am the only user, so by default I have access to this. So, I can basically through UI set these controllers. If I go back to the default project and to go to canister B and do the same here.

[00:13:12]In the future, there will be address books, so it will be a little bit more readable who's what. We already have some naming conventions, but you could basically address save this principal to say, "Hey, this is Quinn's in the CLI." And then everyone has the same notion as you do who has access to what. So, if we try this again, we just do deploy and yeah, successful settings.

[00:13:35]It's installing the canister, it says successfully, then it's syncing the canister again from also mentioned previously in the global R&D. And then we can basically open uh one, two canisters. I made sure I gave canister different colors, but you can see canister A, canister B.

[00:13:51]And so, as a final thing, we went from top-down all the way to the bottom.

[00:13:55]Deploying, we're now live. You can also, of course, now that we're running, we can um stop these canisters. So, um so, we can basically say, "Hey, stop."

[00:14:07]Um you can also see the version increase uh and some metrics data from the UI directly. So, you can see these numbers were zero and now you see there's a total size by uploading basically the assets because we use a an assets canister for this demo behind the scene.

[00:14:20]It has uh says stopped. So, if you go here again and refresh, of course, canister is unavailable.

[00:14:25]And this is also how the subnet admin, so the admin console, can basically manage some uh resources across the whole stack. Uh as a final thing, it's a very nice extra we have and out of the gate, uh again, this is, of course, a subset of the data. It's uh the staging environment. We have very in-depth graphics or graphs uh within Grafana, which will be soon be integrated into the front end and you can basically see over time what happens, who's the canister that takes the most. The peak, I think it's Marcus's canister uh we discovered earlier today. And then, of course, you can also make I quickly made this before the call. You can say, "Hey, just give me the top consumers and their data, of course." Since this is just testing canisters, there's a lot of empty ones we created by testing buttons and different floats. Um So, yeah, this is all from top-down what we've been building on. And so, going forward, we're really taking like the Cloud Engines approach into account. So, this is also applicable and we basically other users can use the console to manage their instances and have an in-depth overview of what's going on on there.

[00:15:26]So yeah, coming back to if you want more information, have questions, just make sure to reach out and that's I think all I wanted to quickly show. I'm not sure how quick I was, but I'll I'm going to keep it at this.