Afri Schoedon - Holešky - a proof-of-stake testnet ahead of mainnet

Transcript

okay well we're going to continue on with today's programming here very pleased to uh introduce afre here who is the head of protocol engineering with chains safe and he's going to be speaking to us about our new favorite test net thank you Ken yeah good morning everyone um I'm not only head of a protocol at chain saafe but but I also do test Nets on the site and um I want to give a presentation today about a new and upcoming test net that might catch your attention um what is a test net

so um the non-technical explanation is if you have a lot of ease and you want to test out a new smart contract you know you don't send all your ease to it and just Euro you know you don't want to see what's happening but you test it you know you set a small amount see if it works before you go in with a full thing so uh in software engineering you do the same thing if you build software you don't want to test it in production right you want to test it before in blockchain systems ethereum we are talking about massive peer-to-peer networks and um in network engineering we need to also test networks right that's what a test net is and today I'm talking about a very specific kind of

test Nets which are the public test Nets um or stable test Nets that are used by a lot of different users to test different use cases who are these users um the users couldn't be more different in in in what they require from a test test net what they do on a test net and what they want to achieve their there are the protocol Engineers that build the lowest layer the actual test net itself or the actual execution layer ex consensus layer in the case of ethereum the huge service providers um there's a huge amount of contract and

application developers that build on top of this entire stack and obviously there are uh all the people engaging in consensus um in case of ethereum it's a stake or staking pools and last but not least the users actually engaging with all these different layers of the test net so um what are the ethereum test Nets um you've probably known um for a long time um there's girly appr Pro of stake test net there's cier approve of authority test net and um you might not have heard about heski yet but I get into this a bit later um uh it's also

proof of stake test net but what's the difference um and why is there a difference um proof of stake obviously is like main net so you know most of you probably stake EAS on mainnet but if you do so you might unless you use a pool where you really trusted you might want to test your setup before doing that and you can only do this if we actually have a test net is also like main net so supporting the full-blown proof of stake and this is uh valuable for two mainly two groups of users um first of

all for protocol Engineers that want to make sure that their algorithm their Beacon chain the consensus layer is actually working and also obviously for everyone who wants to stay Ease on the other hand um proof of authority has been uh very valuable in the past uh to provide test Nets with a certain um stability guarantee so if you want to build applications on a test n you don't necessarily care about if it's proof of stake proof of authority proof of work or whatever or even mocked consensus you don't care you just want to make sure that the application you test and

eventually if you have a larger stti you provide it to your users that want to test it so you want to make sure it's it's also available and uh uh stable throughout uh a long period of time so this is mainly uh valuable for application uh uh application Engineers so when we just launched the heski test net the the question was raised why do we need a new test net we already have two or even more so uh let me just take a look at the motivation here um so the girly test net which is our current proof of

stake test net it has been deprecated earlier this year it might reach end of life it's not really officially announced but we are very close to end of life and the main reason is that we are out of test net eer it's simple as that you know we we if you don't have EA to test you cannot run an application on it right you cannot interact with smart contracts but uh most importantly you can also not uh test staking right um the entire operations of staking you know each Staker needs 32 e and if you have a pool or larger stack of you want to test a fleet of stakers you easily need 10,000 of test net EA to um test your setup and uh since the girly test uh

EA distribution is basically depleted um this created a very um horrible situation of uh speculation so people that didn't get test net EA started to try to buy it on on on on Shady channels so there was a speculation around this asset which should be free for everyone to use so there was a decision made we need to like really Purge this and um just replace it on top of that we also need a better test net because girle was always liking behind a bit so protocol Engineers ideated what is actually the requirement for a new test net and what can we do can we use this chance if we launch a new test net to to um get ahead of main

net oh and last but not least um this would be the first test net we launched after the merge uh last year so we can actually bake in the merge to proof of stake into Genesis so we have a proof of Stak chain right from Genesis which is I believe really exciting this is EF def Ops Team and myself at protocol Beren trying to launch it we didn't manage to but we launch it later relaunch it later um yeah so why is this important what I just said so um for protocol Engineers the idea to have a especially for the consensus layer Engineers to have the idea to be ahead of main net this

basically means to have more valid datas on a test net than we actually have on mainnet so you can see on girly we have currently 600,000 validators on mainnet we have around a million whereas we launched um heski with 1.5 million and this serves like a canary you know if if something goes wrong we want to see it on a test net first and not on Main net and if main net is ahead of a test net there is actually uh the the possibility that we would spot bucks on Main net for us which is not desired so we decided to use this opportunity to go ahead of main net also in addition uh to that we uh

drastically increased the total Circle Lighting Supply sorry the circulating supply for the test net um we baked in almost three billion heski Easter into the Genesis as compared to what the laughable 100 million we launched a girly with also uh we came up with some strategies of Distributing the eer to to users of the test net so we pref funded um 500,000 wallets with, testet Isa on heski so if before you complain on uh social media about getting testet is check your wallet first um shout out to

philli who uh did the heavy lifting there and also um we distributed a lot of um testet EA to important stake stakeholders in ecosystem like e Staker that run their own services to distribute test net e and e staking or testing yeah so uh I think um yeah we just told everyone last year to migrate to girly which was um either a miscommunication or lack of communication I think we were not entirely sure what a strategy for testers would be and um as I said there

are so many different stakeholders in ecosystem and many people have different ideas about test Nets or about stability guarantees and when we told people last year to migrate to girly we didn't really think this through we have to admit this and um that's the moment where we started too late admittedly but we started to talk about strategies out of this mess so a lot of large so if you have a large application stick like I don't know you run huge system of contracts that depend on each other you have new the network effect of integrating with other applications on test net or even run a layer two you know it's a lot of it's a huge stack you Deploy on a test net and then someone tells you you now deployed

on another test net so they lot of projects were not happy about this so uh we um s how can we manage expectations for test net and came up with a predictable predictable testnet life cycle um that has clear timelines and uh predictable end of life expectations so um that um migration paths can be planned long ahead of dication so that projects are kind of aware that a certain test net comes with a certain uptime guarantee and the idea here is to have a guaranteed um testet lifetime of four years and allowing on top of that maybe

long-term support or if if a testet does well we don't necessarily need to deprecate it after four years but we don't really want to commit to more than four years here and um since there are we have both a proof of stake and a proof of authority test net we will launch a new test net every other year if we commit to this plan the the actual proposal we wrote is on eer magicians can uh look it up and this uh timeline just visualizes where we are right now you can see um 2023 girly is at end of life seola is in kind of in the middle now most of the applications

are migrating to seoa currently if you do not migrate yet you might even consider another option now so yeah we also um shut down ring B earlier this year so it's a lot of movement in the testet landscape girly is I don't know what the exact date will be but it's it's about to die we I think we still have to officially announce the heski launch but at this point I would say if you still run girly or ringy or any other test net uh it's it's uh really time to consider migrating to CIA or heski depending on your

needs okay I finish my presentation with a handy test net selection guide um as I said heski is proof of stake so obviously if you're a Staker and annoyed about the girly St um just take a look at uh heski um due to the size and expected Network effects it could be also interesting for large service providers or layer twos or yeah really whoever else needs main net conditions on a test net uh at a time of this presentation know the expected uptime guarantees are around four years with an end of life of 2028 for heski sear um is as I said for whoever

needs stability guarantees mainly application developers and currently the expected lifetime is another two years for this test net repeating the obvious don't use girly please stop using girly painful for me to say this uh girly was a really important project for us when we launched it 5 years ago um but test net changed a lot since uh Al eum changed a lot since so let officially deprecated it stop using it obviously also don't use rink b rubson or Cen and if you didn't listen to me the last

10 minutes this is uh the summary if you want a steak use heski if you want to test application staks use CIA um oh I have a bonus slide uh since I have time how to get test net EA it's a uh it's a Hot Topic has always been um obviously for heski uh sh your testet wallet we did um we did this um airdrop to 500,000 Wallets on heski so um yeah that's

that's that's the first thing you should do uh obviously we have nice forets up the my favorite is the proof of work for by philli um but there are also others um if you want to validate and don't have enough EA to test a valid dator setup obviously eaker is doing a really good job at Distributing or funding your your valy datas directly you can use their Launchpad uh Shan saafe provides a similar thing it's called the valid dat foret so you don't need EA but you can still make deposits and run your own valid datas with their ease and in case you didn't know um if you have tickets on pass you can also claim test net

eer so you can go to the heski foret PK 91. de and uh connect to your Zupas account and you will be eligible to get 500 test net eer I think on every network but yeah you can check it out and that's that's plenty if you want to test staking setups or applications you should be sorted up for a while yeah and last but not least if you really need if you really have a reason for a huge amount of test n EA you can always talk to your local dealers if you know any um that's my uh that's my talk um

that's does there any question oh perfect Excellence questions we'll get right to them thanks for talk uh yeah just a quick question regarding how much is the fish maybe you talked a little bit about it in the beginning but I missed it can you tell us what exactly went wrong with the first testet launch and how it's related to The Fish thank you this is my favorite question uh we have been launching testut for a while and uh I personally believe it's it's a beautiful moment to launch a

testnet because you you create Gus is block by hand and you can like put up a lot of things that remain in the history forever you know um this will be always in the extra data of block block zero for example and uh for heski uh we did not put any and when I reviewed the Genesis blog I I said why not just put something in there and just put the example how much is a fish and um the person who was offering this PO request took it took it literally and put in how much is the fish which is kind of funny today but then it was uh uh yeah it was it was was funny all the

way and um I didn't realize by making this suggestion that I completely um broke the entire testing uh setup or like the the way they because all these files were pregenerated if you change something you can no longer be sure that everything else is fine so what happened is that we we um patched the the Genesis of the execution layer manually without updating the consensus layer uh configuration so um once we launched testet at at protocol Berlin um we uh we didn't manage so the consents layer just refused to talk to the execution layer because they refused the Genesis and

what we had to do is basically just remove this again and launch so it was not a big deal but it was kind of uh sad but also now now I think now we can laugh about it hi afre um since you mentioned the situation with uh with girly um with yeah the speculation with tnet ether um I wonder if there is any solution or any plans to avoid this same situation holis yeah um that's also a very good question so um there are typ uh a lot of things you can do about about this so first of all you can increase the supply that is the easiest thing because if you have more obviously that will last longer

then um also what eaker does or what chains saafe does with the validator forets is basically you put out the the ISA for the people testing but you basically make them to set a withdrawal address to their account so whenever they are done testing it basically goes back to you so it stays in circulation and it's not just because people lose interest or move on they will not just it's not not getting lost right and there are many of these ideas and last but not least we have um on heski we have these mechanisms but since um seoa is proof of authority we have a way to Mint new tokens um through uh through the deposit contract so we

have technically an unlimited Supply there so if there's ever a situation we should be I think we should be good for at least 5 years and just looking at how how long we managed to deal with Sly in The Last 5 Years with just 100 million's total Supply I'm I'm very confident we are set up for future here yeah I have a um last question um can you comment a little bit more on what stability warranties uh does proof of authority offer sorry can you repeat this yeah yeah um you said that proof of authority offers um more stability guant guarantees what are they can you comment

a little a little bit more yeah so in uh before the merge proof of authority has basically you just have I don't know seven validators that run the entire consensus for the for the Shane and if someone goes down you basically just one write one message on I don't know Matrix or and then then ping this operator and you know this person will respond within a couple of hours and then it it's fixed you know so so you the network is not relying on large amount of people or organizations to coordinate to run it like we have on proof of stake but just a couple of few so after the merge the aolia test net is run by let's say 15 or 20 entities that coordinate in a chat

Channel and it's like it's a closed valid dat set so nobody else can join this valid dat set so the stability guarantee here is like on the consensus side right and all other uh stability guarantees are inherited from the main net I think that that's a difference all right um one more question in the back here and uh um I'm just wondering if you can talk briefly about performance and like one of the things that was annoying about

testing validators on girly was that you know You' Miss at a station sometimes and it wasn't necessarily your fault because yeah the whole network is just not performing as well as midet um do you think that over time heski will start to look like that as people are done testing and they just unplug their machines and don't properly exit their validators is there any reason that heski will be any better or worse than than girlyy performance- wise yeah uh I think it's just getting worse the bigger the value said is but what we what we did is we tried to ensure that let me say trusted operators have a super majority in this validator set so it's a lot of all the client teams ISM

foundation and uh long-term um community members are running a huge chunk of these validators to just make sure we have certain finality guarantees for the test net but still keeping it open for everyone else who wants to join so um girly was a bit shaky at some point and uh we cannot rule out that this will not also happen on H at someday but it is it is the nature of of proof of stake on a zero value test net we cannot fully mitigate this unfortunately well excellent thank you very much for a wonderful presentation please join me in giving our speaker a round of applause here thank you