The Impact of Distributed Validators on Staking Resiliency & Decentralization by Brett L. & Chris B.

oh all right hey everyone I'm Chris head of product and this is Brett our head of growth here at Noble Labs here to talk about the impact of distributed validators on ethereum staking just by a show of hands uh who's heard of DBT or distributed validators all right yeah a few of you that's uh that's good to see much more than than last year um and actually proud to be the first uh talk this week on distributed validators I think there's six or seven talks and panels so it's a growing point of discussion so we are opal Labs we're working on um on where where research and software development team working on proof of stake infrastructure for ethereum we're made up of 23 core team members 15 advisors across the world we're building tools to enable trust minimized staking and we're enabling ethereum stickers to permissionlessly run distributed validators so today we'll talk a little bit about what is DVT our design philosophy as well as the progress um our roadmap essentially and what's next so what are distributed validators or what is DVT so DVT is a protocol that allows validators to be split and run amongst multiple machines we call these machines nodes and they operate in a cluster so in the web 2 world or traditional Tech if you have a website or you have servers or apis and they were to your website were to go viral or your servers were to go down we have devops tooling devops tooling to spin those up and make sure that they're up and online well DBT brings that brings highly available software deployments and makes it a standard for validators so this addresses several issues with currently running a validator today you know software inevitably will crash um it doesn't stay up forever Hardware needs to be maintained brought down it's one back up and so DBT allows you to have one node or several nodes within your cluster go down and still be operational online additionally private key compromise and Byzantine Behavior so if your machine were to be compromised or a set of the machines would be compromised in the cluster there's no impact to the validator so how does DVT work um so it starts off with a configuration and then we do it what's called a distributed key generation to split the validator's private key into multiple parts so this year is the year of the spork um so you can imagine if you have like four prongs of the spork you set up like four nodes in the cluster and one of those prongs were to break off you can still run the um you can still use your Spork or you can still run your distributed validator so in the dkg process we'll we'll do the the private key split and uh just three of the four nodes need to come together using threshold BLS cryptography to perform the duties of the validator and it's not just for a three of four you can set up a five of seven or a seven of ten anything in between even slightly larger than that so if you had a 10 size node nodes in your cluster three of them could be down and seven could still keep you operational so who can benefit from DVT uh really just anyone running a ethereum validator today uh liquid staking pools and right now they have to effectively trust uh their operators um with keys and to not be slashed um and so you can set up a your operator Set uh in a cluster and allow for uh one of those nodes to go down or if they needed to eventually remove an operator from the set because they went Byzantine the cluster can do that and continue um can you continue validating online without any interruption uh seeking providers as well can build distributed architecture and they can also set up a multi-org cluster with other staking providers to have a more decentralized solution uh home Solo stakers or at home validators they can build a more resilient validator so they can have their validator at home split into four parts they can put one node in AWS one and another cloud provider one at home one in their home office and if one of those were to go down they're still up and running and uh yeah we've seen a growing use case for communities in Dows to run validators on behalf of their uh their organization and this allows for communal operations so rather than just trusting one member of the Dow to with the funds and running the operations of the validator you can do this in the communal fashion so yeah I just want to highlight that we're not building a liquid sinking pool we're not a staking provider we're building software to enable all these products so if you're staking on ethereum whether it be using one of these other products or doing it yourself you get a net benefit from DBT I'll just touch on our DBT design philosophy so we um we built our DBT implementation which by the way stems from the ethereum foundation research several years ago and we built it in a way that we wanted to make sure that validators today or that we would build a DVD solution that would be more performant or equal to valid errors today with the added benefit of resiliency decentralization and security as we touched on the previous slides and then also want to highlight anti-correlation so we did not build a valid air client and the reason why is that we want to make sure that DBT itself doesn't become a vulnerability so if we just went and built a client and we had like a zero day bug it could actually impact the entire ethereum staking ecosystem uh if let's say we're running 10 20 30 of the stake is running on our DPT implementation and there was a bug there it can actually have a huge impact so because of that we believe that DBT is best designed as a middleware so how this is set up uh is we have Caron it's a golang based DBT middleware implementation and it sits between the valid error client and the consensus client in the ethereum staking stack and this allows you to have to come to the table with in your cluster with actually 100 plus different configurations of of execution clients consensus clients your valid air clients and of course our middleware solution as well and so the middleware client Caron will come to consensus with the other nodes in the network and then pass off the message to the validator client to perform the duties of the validator and uh who here recognizes this little Rhino here um who's seen that before yeah all right so that's Leslie um so a few years ago a couple of us from the team got a chance to work with the ethereum foundation on building the ethereum staking launch pad uh that's still up today maintained by the EF on LaunchPad ethereum.org and um and it ran throughout the test Nets in early yeah 20 2020 and then later on December 2020 um um was uh helped everyone walk through the beacon chain launch setting up their validator so that helps you walk through creating a validator with a single private key and I'm happy to announce that a couple months ago on the anniversary of the beacon chain launch we released our distributed validator launch pad which gets which walks you through the process of creating a distributed validator and the dkg you can actually uh scan that QR code there and that will take you to our deployment for our testnet um so yeah last I'll just touched on before I hand it off to Brett how to create a cluster so you can walk through the UI give it a name um also determine the size of of your um your cluster put in the addresses for all the operators if you're running it as a group and then select the amount of validators your enr which is effectively a way that the other nodes in the network can identify you set your withdrawal address view recipient then you go to the terminal just like as you did with creating a validator with a single private key and run the dkg for your group all right I'll hand it off to Brett to talk about the progress of where we're at in our test sets and and maintenance all right uh thanks everyone so yeah what I wanted to talk about today is where we're at with uh DBT and so DBT has a concept actually started a few years ago and now it's become a critical part of italics ethereum road map and as a middleware implementation I've been working with the entire staking ecosystem to get DBT to where it is today whether it's our 25 000 plus Community uh testing at home validators uh using distributed validator technology or professional validators Liquors taking protocols Etc everybody's really had a hand in getting DBT to this point um one big part of the progress to date is the testing roadmap we have run uh two tests or one test net so far both with at-home validators as well as professional validators kind of in two different tracks that is being combined into a single track going forward to make sure that when we test and develop DBT that it's addressed in the needs of all types of validators in the ecosystem so a little bit about our last test net which was called Athena this ran from July to December of last year and we had a hundred plus distributed validator clusters sped up in 40 plus countries so really seeing participation from across the world and running distributed validators we also partnered with a lot of exiting staking communities like East acre Banks et cetera node Opera operator Association to get uh involvement in the entire community in uh deploying and testing DBT and we've also run Pilots with professional validators like Lido and what I want to do is share a little bit about the performance of the early testing because one of the biggest values of DBT is letting at home uh stakers now run nodes that have similar performance to a lot of professional validators and we actually saw that in the Athena performance test results you could see like the top 10 percent of our validators at home validators saw an uptime of 99 Effectiveness in the top percentiles and so this is really hard to achieve if today if you're staking by yourself now similarly with professional validators like Lido we saw a lot better performance as well in the distributed validators uh setup that they had and so this is again a technology that benefits everybody we are now currently on our second test net called Bia again this is both for at-home validators and professional validators we launched it about three weeks ago uh and the goal was to scale deployments both for at-home validation as well as professional validators and what's been really exciting to see is that in the last three weeks we've signed up over 850 plus at-home validators and this is a distribution of people running four nodes seven nodes ten nodes and it's the thousands of people across the world uh running distributed validator so the energy from the entire community of wanting to participate in a validation um has been proven in this we've also signed up 60 plus professional validators in 14 plus multi-cluster orgs and so you're seeing the likes of different operators professionally collaborating with each other for the first time instead of running validators completely by themselves and this is providing much better resiliency and Safety and Security for the entire network so one really cool thing I wanted to point out is uh and the screenshot's a little bit uh fuzzy but this is our Discord Community uh and really we saw people at home uh organizing around the world just organically through the Discord we open up a channel called be a find a cluster people posted uh a message saying you know I'm a validator I'm located in Asia I want to join a four-person node uh you know here's my motivation for doing so and people just kind of self-organized to create these distributed validators around the world real really exciting to see a very hands-off from the core team at obleton doing this and also this is just some of the leading validators and liquor sticking protocols that are participating in our test net again 60 plus organizations really representative the entire staking ecosystem pushing dbt4 to provide better security and uh better centralization for the entire uh community so I'm sure at this point you have a question when mainnet uh that's a common question I think for for everybody and what I wanted to share today and we announced this actually a couple days ago is we have the first ethereum mainnet distributed validator uh up running today this was launched at the end of December uh last year you can follow on uh Twitter we actually created a specific handle to update the community on what's happening with this Mana distributed validator these are running on nodes purely from homes in three different countries in Ireland Estonia and Canada and again continuing the results that we're seeing from our test net this mainnet validator is performing at the top one percent of all validators as according to rated Network in the last 30 days so really exciting to see the high performance of this uh just double clicking into that if you look at all the key metrics around performance uptime Effectiveness inclusion delay it's performing at or near the top of all of those metrics and so again it's really exciting to see now that at home validators running machines you know sitting in a living room somewhere working together can have really really good performance it's going to improve the inclusivity of who can be a validator in the future so we are about um halfway just from steps wise on our road to mainnet in V1 uh the goal is to launch opal V1 in a broader mainnet deployment uh within within a year um and the phases next we're taking a phased approach to deploying this just to make sure as we increase the pool of who's running distributed validators that everybody's successful and so we're taking you know we have an alpha launch coming up here and then in beta launch later in the year and in between we're doing an attack net just to make sure that from a security standpoint that we have all our bases covered so we're going through audits right now as well so long story short we're getting there uh to a really close to mainnet and that's a happening you know later this year so uh I'm sure many of you are wondering you know how can I get involved um and really if you are interested or if you are a validator today uh join a Discord Community participate in the be a test net whether you're part of a professional validator entity or you're running uh you know nodes by yourself I think this is a critical technology uh primitive to to adopt and to get involved in um so please do so and then also if you're not running validators and maybe you're just staking eth on some kind of service I think it's still important to educate yourself about DBT this is a critical primitive on again the theorem roadmap and it helps to understand it so that you can better secure your stake wherever you're trusting whoever you're trusting your youth with um so there's a lot of materials created bioble by the community around DBT that's just the list right there feel free to check it out and we have four minutes left actually so we can either take questions now or take questions on the side but I'll I'll pause there any questions can we get him a mic or oh hey how's it going good to see you got a question yeah hey all right so uh I use uh you know people like Lido rocketpool coinbase as a provider pay them a fee to run the infrastructure Etc you guys have done a test with Lido and I saw that what kind of incentives exist for people like Lido to adopt oble versus other providers who other players who are competing in the uh in the DVT space aside from the like clear benefit that it provides them less slashing risk Etc but how do you see that from like a you need to compete to get these deals done type of thing and what incentives lie there thank you yeah um so we are working on a potential ecosystem fund on but we haven't like solidified any uh anything firm there yet if I can just touch on like that it benefits to taking protocols or pools as a whole which is that again they can rather than having um you know having to trust their operators to not be evil they can actually just set up a cluster of their operators or a set of their operators uh and voted in through governance and then um if one of the operators were to go Byzantine or even like a subset of those if they have a much larger cluster uh that they can add in the future they'll be able to add and remove those operators from the set so um so yeah I hope that answers the question but nothing firm yet on like direct incentives if that was a the exact question yeah I also wanted to add I think just from a design standpoint uh the fact that we took a middleware approach is very important because it allows staking providers and validators to to create different setups of different client configurations different client combinations and it reduces any kind of single point of failure in their in their Network so uh just from a pure design standpoint you know the fact that we didn't create uh oboe DBT as a validator client or as you know a full stack liquid sticking protocol um the the goal is to help the players in the ecosystem better uh make their networks more resilient more decentralized and stable in the future so all right thanks everyone cool thank you [Music]