Oisin Kyne - The Impact of Distributed Validators on Solo Staking

Transcript

okay hello everybody so my name is Asin I'm CTO and co-founder at oal and today we're going to talk about distributed validators and specifically how they impact solo stickers so what I'm going to talk about is for those of you that are unfamiliar I'm briefly going to touch on distributed validators including some of the architecture decisions we've made as obal when making them then I'm going to get into some of the pain points in Solo staking talk a bit about the kind of history of it and how I got into the space and worrying about Stak centralization then we'll look at how DVS can actually be of use to solo stakers and then lastly and most importantly we're going to talk about how you can get involved today in Squad staking so for those of you that haven't seen any of my stuff before we're going

to talk about what are distributed validators so a distributed validator is an ethereum validator that runs on multiple machines instead of one and it has fa tolerance you can think of it like a multisig for a validator where not everyone has to be online and signing all at the same time this is particularly useful for solo stakers that people that you know don't have perfect up time or you know 24/7 SRE team that can look around and like manage things and what's also particularly special about it is you don't need the same entity running every piece of the validator you can have four different people running four different pieces or seven or more and beyond that and going a little more into the technical details one of

the things that I'm particularly happy with about how we've gone built distributed validators is the idea of making them a middleware which means that Caron our client sits between a validator client and a consensus client and only intercepts the rest traffic between them but does not have runtime access to the private keys I think this is super important for a number of reasons security frankly being top of them if ever there was something that went wrong with the car on client it might show something slash to the downstream validator but the downstream validator has its own slashing database it says I I've already signed something at that slot height I'm not doing it again and you don't get slashed nor do your kind of keys get decrypted NEX filtrated or anything of the sort it's also useful for people and particularly

Enterprises that have their own private key infrastructure it's a really big ask to say hey swap that out and like put in our software instead to say no you just need to add mid middleware to your stack but other than that you keep your own key management system this is kind of a big thing for both you know adoption and again most importantly security what's also important is is you can and should like mix and match all of the different validator and consensus clients within it you should have in like all of your different nodes different Els different CLS different validator clients and if there's a zero day or compromise in any of the pieces of software in that stack so long as it's only a subset of the node that really reduces the risk that something will go very wrong another piece of the puzzle for

those of you that haven't seen it before is what we call the distributed validator Launchpad or the DV Launchpad this you can think of as a successor to the eat staking Launchpad which was an interface developed by the EF and consensus to facilitate the creation of validators the DV Launchpad serves the same purpose and it particularly focuses on helping groups come together to create distributed validators if you run all the like pieces yourself you can just kind of create the keys using the CLI but if you have you know multiple different entities around the world coming together to run these you want to know that you're working with the people you expect so everyone like Signs messages with their well-known like Ledger addresses their ens names you can deploy smart contracts that help divide out their rewards and you can generally

have good confidence that you're running these things with who you expect because if you're running them against like civil attackers or someone that you know you don't trust there's decent reason to be concerned so now I want to talk a little bit about the challenges of solo staking and I want to like lead with the question is taking from home easy yet anybody going to like staunchly defend saying it's easy to run a valid later yes I was suspecting um so I think solo staking is getting much much easier over the like last you know few years but I still think there's a long way to go some of the difficulties of it first and foremost is needing worth of $50,000 to take part um or you can join liquid

staking protocol which is you know another piece of the puzzle the one that kind of calls me is the like being on call 247365 I find very reliably like when not will always die when I leave and I'm out of the country for a few weeks um bandwidth caps matters to some people that don't have kind of unlimited all you can e data I know my uh parents were really unhappy with me when I burned through all the internet in their uh house because I had an uncapped kind of amount of peers running um lots of the world don't have you know the most stable access to power they experience kind of power outages maybe load shedding this can at the very minimum knock your validator offline at a more worst case it can corrupt your database and like force you to like resync Hardware dying is one that for hopefully

you don't get run into all that often but if your you know disc dies and you have to get a new one that's quite a bit of like downtime and outage as you buy another one from Amazon install it sync it and get it up and running version updates not too bad technical expertise is coming a lot down credit to like the work of the node launchers around the space including the DAP nodes Yad sterium e Dockers and so on and even just generally trying to keep kind of high up time high rewards nodes isn't the easiest thing yet but thankfully to a lot of people that are here today it is definitely getting easier so briefly I wanted to talk about where I got into staking and it was actually five years ago yesterday when I

wrote this article but um at the time I had kind of come to grips with the proof of work energy usage and I was like okay this isn't really going to work what is like the alternative people are focusing on and I was introduced this idea of like proof of stake and I was like cool let me you know have a bit of read of it see see what it's about and at the time it was 248 ether minimum um which was about I don't know half a million dollars at the time and slashing was kind of Il defined but implicitly it could have been like all of your Stak could be slashed and with those two things in mind I wrote this article called the staking problem which is effectively this is too hard and too scary and if this isn't your core business model you will Outsource

this to somebody else to do in a professional context and given 5 10 years my fear was that everybody would end up like Outsourcing this to Enterprises and that there would be three main entities that had like total control over the stake and my fear was not necessarily that they would have like stolen money but my fear was they would have done something non-credible neutral begun to censor begun to kind of maliciously like weaponized Forks against people and that was kind of my concern was taking at the time and I was frankly a little bearish and maybe even suggesting we could like you know think a bit longer however over the course of the years the technology improved a lot one of them being the adoption of BLS signatures allowed the stake to come down to 2048 ether and there was a talk at Devcon 5 in Osaka by these two

gentlemen from the ethereum foundation Carl and donrad um with the help of a DTA as well where knowing that BLS signature schemes were homomorphically additive they were trying to tell the world there is a better type of validator we can make now we can make a validator that runs like a multisig as we use the words now this is kind of the Foundation of what distributed validator technology is and I actually didn't see this live I would have seen it in early 2020 while I was building block demons e two staking and I was immediately a convert I was like perfect highly available validators validators that run on multiple machines validators that can be run by multiple people this is definitely going to be a thing and 5 years later or three years from this

that's kind of you know where we're at I'm super chuff that through all of like the showings today and all the different schedules there's so many teams talking about DVT for this that Squad staking and so on so that brings me to why DVS particularly for solo stakers I can understand I spent a lot of time pitching it to like the Enterprise side of things but maybe not as much um talking about in the Solar sticker side so the first and foremost one for me is the idea that you can have you know normal solo Staker level of like time dedication up time but end up with you know 39 are nearly Immaculate uptime due to the fact that you have a group and so long as you know three out of four of you are online or five out of seven you of you are online your validator will be

online this allows you to kind of accept going on holidays and having a node breaking and fixing when you come back or not having a totally ready to go second staking machine synced and like ready for a failover and not you know paging yourself in the middle of the night when your validator goes offline things like that will generally allow solo stakers to be indifferent from like Enterprise performance and that's one of the big things that I think is important for moving more of the stake onto like solo sters and away from kind of you know the cloud frankly um this idea of lowering maintenance overhead is kind of what I'm talking about with the discs I got caught a while ago where I did not prune my disc in time and ended up filling it up to the brim and then couldn't prune

because pruning needs disc space so my only option was to like Nuke the disc and like resync the full node um that would would have been you know a couple days of downtime hasn't I been solo saking but this node was actually part of a distributed validator cluster so it didn't miss like a single attestation which was very handy um one of the big ones which I'm going to talk about a little more in some of the coming up slides is the idea that with distributed validators it becomes more acceptable to have solo stakers take part in liquid staking protocols because you don't need to trust them the same way you do when they have like total control over the validator key um I'll dig into that a little bit more as we talk about ether 5 and operation solo Staker and Lido simple DVT module and the last one on

this is related to the kind of collateralized liquid staking protocols the kind of Rocket pools and others at the moment they're all doing lots of good work trying to take their bonds down safely but without exposing their protocol to risk a ether is still frankly a huge amount of money for a lot of the world especially for like young people if you can instead have four people all together put in two ether and run a mini pool with a group that is you know another way for you to kind of continue to bring down these collateralization requirements without you know risking the protocol itself you just say let's all team up and do so so how can you actually take part one of the things that I'm delighted to announce and we announced it just for the last while is that we are now in

open beta and we're encouraging everybody that wants to and has the means to to run one validator on mainnet and we've still keeping kept at one validator for the time being and we have a couple extra pieces of the puzzle that I want to draw your attention to so the DV Launchpad I've already talked about that's um the user interface for setting up validators but now as we kind of move into beta and away from alpha we're starting to focus more on developers of liquid staking protocols and just general staking Solutions and the two pieces that might be of interest to you guys are the SDK that we've now shipped which is a typescript SDK to provision distributed validators and to retrieve the deposit data after key generation

ceremonies this allows you to you know hide the use of DVS behind your user interface make it kind of like a native part of your flow and the other piece that I'm probably nearly more bullish on is what we call obal splits these are smart contracts based on the splits. org teamw work which allows you to without any upgradeability any mutability and without any Oracle split ethereum validator rewards deterministically across a number of operators and you know the protocol itself so this is something that um I've been working on when I tried to make nft validators in about 2020 and it wasn't really feasible until the like post merge and we've

recently completed with an audit with it that you can see in our website but I'd highly recommend if you know the smart contract side of this is what you're interested in um these things can scale to thousands of validators and there's no editability or no trust assumptions by your usual smart contract risk and last but not least karon our middleware client is going from strength to strength we're just beginning a second audit of it now and we're continuing to add features and improve its resilience one of the next ones we're working on is partial deposits to make it easier to deposit into something like rocket pool and the likes so from December 1st what you can do is you can team up with other people find a squad I highly recommend you know

finding them today while you have all of the Solo stakers in the ecosystem around you can go onto the beta Launchpad or I would strongly recommend do this on whole scare gurly if you haven't run a main validator before please you know be sensible um you can you know set up your distributed validator pick the withdrawal address fee recipient so on you do your distributed key generation ceremony make backups of your keys turn on your notes make sure they're connected all you know all the noes synced everyone's ready to go and then activate your deposit and activate the validator we are also working on a credentialing kind of effort at the moment where we'll be issuing non-transferable nfts to anybody that runs valid dator successfully on mainnet or test net because there's a lot of

liquid staking protocols in the space that want more solo stakers but trying to figure out how to kind of due diligence them and understand whether or not they are like credible or not has been kind of difficult so our step in that direction is to just drop nfts to people that have run validators so you can kind of look at them be like yeah they're holding a test net and a main net you know Technic credential they probably know how to run a DV the next thing I want to talk about in terms of ways to get involved particularly on mainnet is with ether's operation solo Staker operation solo Staker is an initiative started by the team a number of months ago in participation with dno aado as well and the idea behind it is ethery wanted to have solo stakers running their nodes

from the very beginning so they did an kind of an open call for applications received a hugee amount of Interest made it like a selection of some of the early adopters and sent them dap node Nado machines to their homes helped them get them set up established running DV clusters and then be began activating validators on them I think at this point they now have 40 solo stakers from 35 countries around the world taking part and challeng aridian I think the plan is to scale this up even more in the near term so go find them if you want to get involved the other main initiative on Main net for distributed validators is on the liof front Lio have recently um passed a proposal to put half a percent of their state St on distributed

validators as part of the simple DVT module simple DVT will be their first foray into running distributed validators on mainnet and it will also be the first time that solo stickers will take part in Lio as an curated operator they currently we're doing a kind of a dress rehearsal on hsky with them as like a final kind of practice before go live there are 212 operators taking part in oal clusters and if all of them were to go to main net that would be the guts of a 500% increase Li of operator set so I'm really bullish about that and if you can find will Shannon around he'll be talking about simple DVT I believe and yeah that should hopefully be on in Maye of this year or mayet early in the new year and they will be doing multiple like waves

of on boardings so if you do want to get involved there should be an application form somewhere in the internet and one of the last things to touch on is the dev side these you know ether and Lio are just two examples but there's a lot more liquid staking and just staking projects gener generally out there um i' strongly encourage you to have a look at the SDK and the splits use them together um I think they make great building blocks that are low trust and you know helpful to allow distributed validators to be incorporated into your product and the idea being um you deploy the solidity set up the Splitters as you so choose create the distributed validator that exit to set Splitters and then you get your node operators to run create the

keys and then activate the validators um the last piece of the puzzle on the development side is we're putting a lot of focus in getting distributed validators built into all of the public good Community launchers so that's your kind of eth Docker your sterium sge um D Nado obviously and more we're super Keen to have DVS to be you know the easiest type of validator you can launch so if you're involved with or working in any of those pack like packages or products would' love to chat to you about it afterwards and one of the last things I want to talk about is kind of the community side of staking so Squad staking is this idea that you run with your squad you run with people you know your friends people that will like back you up when you're not as offline and when your like when theirs are offline

you got their back um La yesterday we were at the shifi summit and shifi is a group of like women in crypto that train each other how to like you know use defi use staking and so on their current cohort is ongoing they've about 400 people like learning at the moment and oall will be doing the staking week where we will get them all to be doing distributed key generation and creating validators and my goal between you know shifi and all the other communities e Staker being a great example is to have loads of people able to set up DVS together we have workshops going on there was one just half an hour ago here um aridian is doing one with ethery I believe there will be many more workshops going on but my goal is that you and your you know three friends five

friends whatever can set up a distributed validator and can seek delegation you you can run your own stake wise Vault you can put put it into node set you can apply for Lio all of these things is generally how I want to see the kind of staking Market mature and not be in a like a world where it's all just like big Enterprises but it's instead people like sticking with their friends so yeah that's more or less the thing I want to leave you with guys is it's time to make Squad sticking a reality give like run nodes if you can develop with them if you can and would love to get more people taking part in eat proof of State consensus that's me thank you very much everyone we might do a few questions [Applause] thank you oen that was awesome uh any

questions from the audience just a really simple practical question to start around um uh like what we often call T and N are is this configurable at all what I mean the threshold the number and the threshold is it configurable in any way like I can have two uh I need two signers out of 15 or is there certain requirements there and how should one think about these trade-offs uh you could we generally would not recommend changing it off of what what the default is so the default is strictly like 2/3 or or specially 2/3 plus one of the number because there's a consensus mechanism involved and if for

example you had fewer than that let's say you had like yeah a 7even no cluster but only two to sign two different groups could like sign something going be slash but instead it's like if you keep five as the threshold and your consensus works there shouldn't be a scenario where you sign two slash messages so yes it actually is there in an advanced flag if you want to go change it but it's probably safest to use the default excuse me so obviously the Caren middleware is running alongside the the usual clients hello hello hello hello hello hello hello um yeah so Karen is running alongside the usual client so it's like yet another piece to this you know stack of

of clients that ever seems to be growing you know me boost and two clients and now Karen as well so is there increased Hardware requirements if you're trying to run obal at home or what's this situation good question um the increase is pretty negligible caran will use um 60 to 100 megabytes of ram um so if you are not cutting it extremely fine you should be able to include that and it doesn't really have any disc um it just has its config file so a couple kilobytes of disc so the the main thing maybe is the like peer-to-peer communication but again that's kilobytes a second so nothing relative to actually running a beacon node any more

questions all righty well thank you guys for coming to this talk the next talk in this room is actually canceled so we're going to resume here with some more talk about DVT and the D.A team at 24 let's give it up for oen one more time thank you very much everybody