Minimum Viable Data Availability for Optimistic L2s

Transcript

foreign [Music] hello hello is anybody here from optimism raise your hand you we need to talk all right hello everyone super happy to be here with you in Paris I'm going to talk about minimum viable data availability on optimistic l2s my name is Jerome I run a company called cometh and we have a super cool middleware street called alembik not Mozambique um and we we run Onsen games that's uh

that's our Core Business we we like to do two games and we'd like to do good Tech four games uh and also focus on retail applications that coming from web 2 to web3 so this talk I'm wanting to talk to you about um three things first that when you want to do Innovative stuff on blockchain it starts to be super gas heavy pretty fast because you're not relying on any standard contract you are trying to innovate you are trying to move fast and break things l2s for this can be great but they're still too expensive and we are going to take a little deep dive on on this topic specifically we want to propose also solutions to solve this in the very short term and eventually if we still have time we're going to talk about our implementation within the op stack of an

interior or user experience Innovation Centric L2 so three years ago in 2021 February we started this first auction game one of the first of its kind along with dark Forest roughly at the same time it was a real slime strategy game which I gave a presentation at hcc4 and a live onstage demo this game still runs today even though we don't create more content for this game at the moment you can still play it because it's living inside a blockchain called polygon in February 21 when we launched it was super cheap doing an action in this game costed as one tenth of a sense and it went from one tenth of a cent to 25 cents 30 cents

to do something in the game the reason why it went so high in gas price and in price for the users was that the native token magic pumped and the average gas price went higher as well along the line well we we had users that came from directly digital ecosystem the polygon ecosystem and different um evm based ecosystem we've already avoided already something and we wanted to try to onboard people from outside the web3 world so we did a lot of things to try to onboard them uh we had a bad experience with funding of Acquisitions we were paying for campaigns and realizing that instantly on what it was tough buying crypto was tough everything that JCP for me explained all of those things

that you have to do in before getting into a cool app so we started to do a lot of research on how we can lower this cost on L2 notably and also how we could use icant obstruction on this and my CTO will give a talk specifically on account abstraction tomorrow if you if you want to get a better sense of it so our conclusion during those tryouts those those three years of Trials regarding onboarding people from web 2 to web 3 and building Innovative stuff like games was that when you have has when you have high gas prices it's often time because you have lots of cool applications regarding D5 and once you have those D5 applications that are everywhere around your blog space it starts to enter The Innovation because it's getting more and more expensive for people to build something on testnet

deploy against test and also I can abstraction is really really something that can solve almost every user acquisition frictions that we have right now but it's still limited to the cost of the execution layer we have underneath talking about the cost um so you've seen this kind of comparison with level zero level one level two Roll-Ups and kind of nomenclature in a phyton world so the way is a roll-up which is not a warlock what is L2 but it's not L2 and so on and so forth let's look at the the cost metric here optimism and arbitrum are two very uh very good roll up very very popular if you want to Mint 1 million ft or say otherwise if you want to spend

111 115 billion gas it will cost you roughly 21 000 to spend that's kind of expensive though um and you know you can you cannot really compare gas expenditure from one chain to the other from one transaction to the others because there are many moving pieces into that that's a good metric to have in mind minting 1 million nft is 21 000. at July the 11th situation of gas price and each on price so we had many talks today about Core Concepts around the the Roll-Ups so I'm not going to spend too much time on this

I assume that you've been seeing this before uh I need a a interim Centric roll-up works with a data availability that's settle on interior so you write everything on ethereum the consensus come from it around the settlement goes on ethereum and the execution stays on the roll-up that's the core concept so practically speaking these three of transaction here happens on the rollup I executed on the roll up and then from time to time you have a transaction on the L1 that takes all of those transactions compress them a little bit and update a roll-up contract by saying the previous route of the state was that and now the post the new route is that and the transition between these previous route to the next route is all

of those transactions for reference you can read in French the translation and incomplete Guide to World by vitalik around interimfriends.com so let's focus a little bit on one of the core driver for cost data availability those are hard questions that data availability has to answer where is the history of what happens taught yeah where is it uh is this place safe is this place a a trusted one is displaced a good one and how available is this data like if I want a sample can I get one can I get it anytime like what is the latency to get me a a sample so I can verify that things are going all right and also how does putting those data in this L2 were installed or this chain where you store or this solution

were installed how does it impact the cost of executing on level 2. and eventually also what happened if everything goes down what happened if the sequencer goes down whatever will happen if all the RPC nodes are not responding anymore what happened if you have to prove that the transaction was malicious what happened how can I Rely to the L2 for this how can I Rely to the data availability for this as well so in the optimism stack you have the difference between L2 execution fees and L1 data fee that's what you pay when you pay a transaction on optimism and it's a similar concept that happened on arbitrary and also optimistic chain when you pay your transaction on an optimistic workup you say what is the current state of the L2 is it a is there

any is there a lot of transaction is it crowded is it not crowded and also how crowded and how active is the L1 and most of the costs come from the L1 data Fe it turns out that uh uh Yoku from outlayer has covered that in extenso in his Talk of the 17th so check it out I'm not going to dive too much into this but today you save very few on the data cost and you save a lot on the execution cost we have a plan at the Eaton protocol level to introduce eip-4844 and downshotting to greatly reduce this cost of L1 data and L2 data as general but this Improvement cannot come soon enough

let's say we have a rollup called roll Maps it's a nigerian-centric roll up so it's putting all of his data history on the L1 it uses the L1 as a DA assume roll Maps consume two percent of the L1 gas which is roughly what some of the biggest optimistic Right Now consume how many copycats with the same time effectivity of world maps can be put on the L1 without EAP 4844 handling some of the throughputs well very basic answer is like okay there's 100 it's consumed two percent we can do 50 of them well it does moving pieces like some something will adjust itself maybe there's optimization here and there but still soon enough if we don't get good

Solutions on the data availability l2s are going to eat the L1 they're also kind of bloating the state at some in some sense so we should pay a lot of attention that if we cannot get down sharding an EAP 4844 soon enough we will have to take decisions and be more um be be more aware of what we consume in terms of data so another term of the L2 nomenclature is optimistic chain when we use data availability of chain and off chain means not on ethereum not directly on the L1 so comparing a basic e-term roll-up Centric you have the data stored everything lightly compressed is put on

the L1 in case of optimistic chain you can go down to posting fixed size data pointer saying I'm doing my data availability there here's something you can verify and here's what you should read what you should read and what you should find so this concept of data availability committee the persons keeping the data in case of a basic ethereum roll up it's literally everyone every node if you run a node you're keeping the data optimism if you run a node you're keeping the data of Amber Trump if you run a node you're not technically keeping data of an optimistic chain because they are more mindful of where they store the data eventually after EIP 4844 the data availability committee changed a little bit and we can all Converge on not

keeping the data on ethereum please don't load the state how long how low can we go we have a minimum security trade-off so we tried for the past six months is to overclock this we'll work on the optimism Bedrock stack and try to integrate a DAC data availability committee inside the optimism bedrock and we had a different approach on this different reference to what is the general philosophy of a blockchain what have we heard what have we learned and learned over the years doing blockchain stuff so with that availability here we try to make it so that no one's really control them like a blockchain that's eventually the infrastructure has no

control point of failure that's a good data variability and also like when we query this data availability layer we should get a consistent response like there is a commonly agreed state of what's the data currently and you know in a movement of making sure that this data availability layer can be decentralized and can be um passed over to the to a community of of users anyone willing to take part in this data availability committee should be able to do so with reasonably low capital and and Technical effort and time efforts so we ended up with this and we're proud to introduce to you that uh we kind of overclocked the thing uh so we are about to release a new L2 focused on user experiences and super low transaction

fees called Master as in we should all get together and our Benchmark for this layer is that we can spill 115 billion gas and one million means for about 400 consistently so that's a huge Improvement and we are super glad to share that with you today we would love to tell you more about it so here's the Min maxing approach that we had regarding where is the L2 stored the L2 data stored we are looking at several different solutions it's very modular and very easy to modify and improve we are notably looking at Solutions like swarm where swarm was the initial

storage layer of ethereum in the early days of the interim protocol and it turns out it can be used for some of the data availability problems that we face today and swarm is 5000 nodes so so kind of a decentralized network as its own the data itself if the data vdbd committee is done right is very good it can be stayed in there forever it can be replaced potentially always like it's a good data availability layer because you have Solutions made for that um when it comes to impacts on the L2 it's kind of minimum it depends on how uh how extreme you are into keeping the data how many replication you want but it can be really minimum and still very available and If the DAC is online but

the L2 goes wrong well it's as good or as bad as optimism depending if you see the bottle are full of or of empty so if you're interested to talk to us about that you can meet us at our boost we have one upstairs if you want to hop into the test net it's up and running you can try it if you want to get a deeper dive on the code we're happy to share and give you access and let you review for us to open source these code and get community reviewed we needed to have a couple of PR merged on the optimism on the vitibism side and it turns out three hours ago they started to merge it so thank you optimism we are very happy to be part of the community now

um well that's all right that's uh that's what I had to share with you super happy to have you in Paris again and I think we have one more minute to take some questions if you have some any questions all right then see you later thank you so much [Applause]