Rewiring the Rollup Stack: Modularity, Composability, and the Road Ahead

EthCC[8] · 1 Jul 2025 · 15:34

Polina Aladina examines modular rollup architecture, from sequencing to data availability. The presentation discusses emerging designs and the role of shared sequencing networks such as Espresso.

Speakers

Topics

From EthCC[8], hosted by Ethereum France.

Transcript

And we are rolling. Yes. All right. Thanks everyone to come visit me at my talk. Uh today I'll be talking about how do we rewire the roll-up tech modularity composibility and the road ahead. I'll be staring at my laptop because I can't really see the screen. Uh all right. My name is Palina. I work at Espresso Systems as of recently as a TPM, technical program manager. And today we'll be talking about what's Espresso, what we're doing, why how did we end up here? And so a little bit before that, uh Ellie, who is also from Espresso, did a more technical work walk

through about uh sequencer confirmation, how do they work? And if you want a deeper dive into how it actually works, how interrupt works with espresso, you should check out her talk which probably was also streamed so should be available in YouTube. So this talk will be mostly exploratory and less technical on what is espresso and why are we here and yeah the main takeaway from Ellie Ellie's talk is that we need fast and secure confirmations to solve the interrupt. So uh how modularity came to be here. Let's start with uh roll-up ecosystem growth and this is a compilation by perplexity of when the roll-ups came to be. So

those are dates I think are approximate and we can clearly see that in like 2019 18 and early 20s we had a lot of rollups show up show up on the scene and after that as a reaction from Ethereum in October 2020 Vitalik Bhuterian published a pivotal post titled a rollup ccentric Ethereum road map and you this is from Ethereum additions and you can see the date October 2020 so and it's marked as a major turning point in Ethereum evolution And rather than fully pushing their original complex share execution road map for Ethereum 2.0, Ethereum has kind of switched uh to let rollups do the execution. So the execution goes to

roll-ups and Ethereum kind of recent to this approach phase 1.1 and done which is use Ethereum namely for consensus and data availability so that layer 2 roll-ups scale uh handle the scalability and uh basically on that moment of time which was 2020 you can already see that roll-ups have achieved like 10 to 100x scaling and basically what they needed at that moment of time was efficient DA day. So, Ethereum kind of pivoted to the road map that would allow uh better DA and this is a picture from the internet where you can see kind of how many roll-ups we had in 2023 and how many roll-ups we had at 2024.

And uh yeah, with optimizing for scalability and privacy, we basically have two groups of roll-ups. optimistic roll-ups which assume correctness and rely on fraud proofs and we have ZK roll-ups which prove correctness up front where Zero knowledge proofs and it worked. So dozen of L2s emerged innovating in various directions while Ethereum solely concentrated on consensus and data availability and with so many rollups the basically next problem which kind of became a bottleneck uh is the data availability because roll-ups needed to publish their transactional data somewhere trustlessly so that anyone can reconstruct the state

and verify proofs. So initially all the rollups used Ethereum call data but later switched to blobs with Dun upgrade which was in March 2024 and then you have different other data availabilities uh protocols available like Espresso uh Celestia veil IDA and basically when you decide on data availability you have trade-offs in between all of them so you have cost security integration complexity so for example why someone would probably not want to use Ethereum because uh it might be pricey and uh while sec Ethereum offers bigger security um yeah the result is multiDA world

where rollups can mix and match based on their needs just like choosing like a cloud provider do you want AWS Azure or something else and in crypto anyone can build anything so uh starting with DA and Ethereum has many different other layers for example execution abstraction, proven sequencing and interoperability, storage, uh data availability tooling privacy oracles. We also have RAS providers which is rollup as a service. If you want to build your rollup, you can basically come to RAS provider and select the blocks that you want your rollup to be built with. And uh it's easy to spend spend any rollup for your application right now and that is possible because

BFT consensus and BFT consensus a is a foundational element for so many systems and such a universal tool that you can solve pretty much any problem with BFT consensus. So most of the uh building blocks that you see here are actually are on top of BFT. Uh so with all of that the modular vision became real and now execution is handled by rollups. Consensus and livveness comes from Ethereum. Data availability is provided by third party DA layers like Espresso or Celestia Veil. Confirmations you can choose Espresso because we have consensus protocol for that. uh and apps can rely on RAS to help with infrastructure.

And while this all sounds good, while rollups have scaled Ethereum, it also came with some trade-offs because we have so many small layers now. Isolated execution environment means a lot of asynchronous composibility. So the function calls in between apps on different rollups can only be included in the same block when they were all on L1. Now if they're on the different layers you have to go through L1 usually Ethereum to so it's kind of asynchronous you cannot do it synchronously and yeah uh bridging is slow because you have to pass messages via the L1 usually 15 minute finality and fragmentation of liquidity users and devs because rollups kind of become enclosed within each

other and it's hard to bridge anything from one layer to another and the result here is broken composibility as mentioned different latency trust assumptions Crosschain UX is clunky and developing applications is kind of cumbersome because of how many like bricks you have underneath the app. And this leads us to this sort of problem. What's your name? Tony, what's your name? Ezekiel. And it's kind of shows the light latency of communication. And uh I I really wanted to include uh this picture here because it's very very important from the perspective like we we will be looking at the road ahead and this is a road and you can kind of see like if you know roundabouts are

actually a very good solution for cities because they slow down the traffic and they usually work with low amounts of traffic but when you try to scale it just putting more lanes in it just does not work and it result like this. So you can see there's like a lot of inputs but there's very little outputs because they're blocked here and uh yeah so this is something like uh settling on L1 can look but uh to have it a little bit more transparent and easier to understand how it works how a synchronous composibility works we can see like this would be the layer one and this would be layer two. So if you're like somewhere here and you want to go to another building, you need to go down and then come here and then you go here.

So if you ever lived in multi-story building, you know that it would take 10 minutes just to wait for elevator, then get out, then wait for second elevator. So it's the same it works with uh layer one and layer twos. So and here we're coming into all you need is espresso and how we're working with composibility and how we're targeting this. So all you need is espresso and how L2s usually use espresso. This is a very nice infographic that we have on our website. So on top we have transactions those small dots that user submits to uh their roll-ups. Uh then they get sequenced in the rollup and espresso confirmation layer. Here we have all the transactions

from different rollups. So this is like a block and this is transactions and then we mine them in here and we provide proofs that this is settled and then we settle that along to L1. And idea here is once the sequencer submits transactions to Espresso, Espresso gives validity proofs that it will not go back. So once it is here you kind of usually you have proofs from sequencer which is soft proof when it's landed on L1 and then you only have validity proofs a little bit later. So Espresso works at the same time as the soft

proofs from sequencer. So uh long story short we're fast. If you need fast secure confirmations uh use Espresso. Uh okay so Espresso provides what rollup needs needs but never has fast secure finality confirmations in seconds not minutes because our consensus is super fast universal compatibility we work across rollup stacks VMs and even different settlement layers we have secure consensus hot shot is like security guarantees without the weight it works pretty fast and crosschain composibility because we have very fast confirmation s you can do synchronous composibility in between chains because different blocks

from different chains they end up in one block in espresso so we can kind of verify transactions at the same time so it becomes synchronous so this is not the hero we deserve but the hero we needed and going back to my remark with um roundabouts so basically the best road intersection possible for the highways is a pinavia. It's not really implemented anywhere in real life, but it's a concept that everyone talks about and people build it in city skyline. So, this is me trying to do some chat GPT images with Pinavia. But the main thing is when you have traffic and you go for example from here to there, you don't

need to decrease the speed, you don't need to stop, you just go where you need to be and you do it fast because you don't really have any like latency for making decision. You just you just go you don't have to stop. And that's the point like we have ultimate composibility and we have Espresso that provides the infrastructure that lets everyone to go fast from where they need to go to where they need to go and you don't have any delays and you don't have roadblocks. You just you just go. This is Espresso. All right. And currently we're on mainet zero and it marks a major milestone in Espresso's path towards a modular composable blockchain future. And uh global confirmation layer is now live and it supports crosschain

coordination. Rollups can benefit from fast bridging and crosschain synchronization. And the hot shot consensus protocol debuts in production with a robust initial validator set. Network is built to scale. So currently we have one megabyte blocks. We we want to scale it to 5 megabyte. Currently we have a rough confirmation time 8 seconds but we do target subsecond finality soon. And we are focused on decentralization and economic security via proof of stake. So it's an ecosystem and uh the more people join us the better benefits for everyone. We are launching our proof ofstake mainet one soon later this year uh hopefully Q3

will it will represent a critical step toward our mission to a permissionless trust minimize confirmation layer. Currently we're permissioned but with this release it's going to be permissionless so anyone can join and yeah it has development benefits will have better uh DA so VAD improvements have demonstrated substantial performance gains means the DA will be faster and yeah reduce confirmation times of on deaf for approximately 30%. All right, join Espresso ecosystem and all if if you're a rollup, come work with us. You'll have faster confirmations and composability. So you'll be able to do some interop with

other chains that are in Espresso. If you're a builder, come work with us. You can build an app that would have a seamless US across all our roll-up stack. If you're a bridge, we can still utilize fast confirmations and crosschain communication. And if you're a node operator, surely uh you'll contribute to consensus and our data available uh availability. All right. Uh our Twitter and documents and that's it. Thank you so much. And if you want to talk to me or contact me, I have some of my contact data here if you have any questions.

Yay. One question. Hey, great talk. Um, I guess people who usually complain about the fragmentation, especially in Ethereum, um, I would say like 80% of the time they haven't used stuff like across and now Espresso is like shipping things which are like even better than that in my opinion. Um, what we really have is like under the until the core protocol can address some of these fragmentation issues in a more native way with native bridges or whatever it is. Um, there are alternative options out there that like the wallets and whatnot need to integrate. And so I kind of see this as Espresso's big challenge is how do you incorporate this into all the various

frontends that people are using? What is your strategy? What are your thoughts there? Oh yeah, good question. Uh, first thought is have a good partnership department. They're doing great job. Uh, but I think it's kind of the same as with account obstruction. If you have some sort of standard and you have like top three to five platforms that use it and everyone has great time with it, that's a good way to go. So basically uh in crypto I think if you have like top three players in the market, most probably everyone will follow uh as long as users are happy. So I think the fast confirmation will definitely benefit everyone. So it's just a matter of having adoption from uh yeah from the top three players on the market which

we're working in or working on. Yeah. Yay. Thanks everyone for coming here and listening to my talk. That was almost my stressful