读懂「DA竞赛」：Celestia、EigenDA和Avail，谁会是最终赢家？
Original title: On data availability layers
Original author: Bridget Harris
Original compilation: kaori, BlockBeats
The data availability layer has become an important part of the modular architecture, acting as a pluggable component to reduce costs and scale the blockchain. The core function of the DA layer is to ensure that on-chain data is available and accessible to all network participants. Historically, each node had to download all transaction data to verify that the data was available—a highly inefficient and costly task. This is how most blockchains currently work, and is a barrier to scalability, as the amount of data required for verification increases linearly with block size. End users suffer here: data availability costs account for 90% of the transaction costs incurred by users transacting on Rollup (it currently costs Rollup $1300-1600/mb to send transaction data to Ethereum).
The introduction of Data Availability Sampling (DAS) fundamentally changes this architecture. Through DAS, light nodes can confirm that data is available by participating in multiple rounds of random sampling of block data without having to download each entire block. Once multiple rounds of sampling are completed, and a certain confidence threshold that the data is available is reached, the rest of the transaction process can proceed safely. In this way, the chain can expand its block size while maintaining simple data availability verification. Significant cost savings are also achieved: these emerging layers can reduce DA costs by up to 99%.
A very apt analogy for DA in 0x ngmi
In addition to achieving higher throughput, a data availability layer also makes sense to improve interoperability. Cheap DA will inevitably drive a Cambrian explosion of new custom rollup chains, making deployment increasingly easier through rollup-as-a-service providers such as Caldera, AltLayer and Conduit. However, as L2 and L3 ecosystems emerge, they become fragmented by default. Getting users to use new platforms is already difficult—it gets even worse if interoperability, liquidity, and network effects are limited. A unified DA layer as the foundation of each network will make the flow of funds simpler and attract a wider user base.
Caldera and other RaaS providers will enable projects to choose a DA layer when building custom aggregations
Avail, EigenDA and Celestia are the protagonists in the DA ecosystem - each serving the same space but taking slightly different approaches to infrastructure stack, execution and go-to-market.
In terms of technical architecture, Avail, Ethereum, and EigenDA adopt KZG commitments, while Celestia uses fraud proofs to confirm the correctness of block encoding. Generating KZG proofs, while a very strict approach to data availability, imposes more computational overhead on block producers, especially as block sizes increase. Celestia, on the other hand, assumes that data can be obtained implicitly through its fraud prevention scheme. In exchange for unfinished computational work, the system must wait for a fraud-proof dispute period before nodes can confirm that the block has been accurately encoded. Both KZG proofs and fraud proofs are undergoing rapid technological advancement; their trade-offs are likely to become more complex, and it is unclear which mechanism will be strictly superior to the other.
For Avail, they adopted the KZG promised architecture, making it a perfect fit for zk structures. This could pose a challenge to Celestia if zk becomes dominant in the future and Celestia relies on optimistic fraud proofs. Additionally, Avails P2P light client network is able to support the network even if all full nodes are down; in Celestias architecture, light clients cannot run without full nodes. Both Avail and Celestia use erasure coding under DAS (distributed storage) to break data into fragments, adding redundancy and allowing the data to be reconstructed for verification.
Compared to Celestia and Avails technology stacks, EigenDA fully leverages Ethereums existing infrastructure. If data needs to be sent to the Rollup contract to prove its availability, EigenDA will inherit the same finalization time as Ethereum. If Rollup fully adopted EigenLayer, it would be finalized much more quickly.
To achieve consensus, Avail adopts BABE + GRANDPA inherited from Polkadots SDK, while using Nominated Proof of Stake (NPoS). NPoS is used to nominate a group of validators that delegators want to see elected, while BABE dictates who will propose the next block and GRANDPA acts as the block finalization algorithm.
Celestia uses Tendermint as the consensus mechanism, allowing users to stake their TIA to receive validator staking rewards. Although Celestia is able to achieve fast determinism through Tendermint, due to its optimistic architecture, there is a waiting period for the guarantee of actual data availability (users must have time to submit fraud proofs).
EigenDA itself has no consensus, but has two mechanisms to ensure the validity of data availability:
Proof of custody: This is essentially an economic security mechanism that ensures nodes store data, but does not actually guarantee that the data is available to everyone in the network. Nodes will be slapped if they dont comply, for example if they cant prove they own the data.
Sufficient decentralization: Ensuring that the set of operators remains decentralized and collusion-resistant is critical to the proper functioning of the network. With a large and independent set of validators, the provision of data becomes a competition with many market participants willing to join. At this scale, collusion is extremely difficult.
An interesting point worth mentioning is that Celestia’s active validator set consists of the top 100 validators with the most staked tokens, and this threshold may be lowered in the future. Additionally, each of their validators stores the entire dataset. EigenDA will optimize each node to store a small portion of the data (potentially millions in the future), so if enough nodes are honest, the data can be reconstructed. The full origins of EigenDA (and more details) can be found in Sreerams recent post.
Finally, Avail provides a useful comparison of the core components of the main DA layers.
New discussions have also emerged about the trade-offs of each design. David Hoffman pointed out that Celestia is a complete blockchain in itself - a complex stack that requires more than pure DA. EigenDA, on the other hand, is just a set of smart contracts, but it relies on Ethereum, while Celestia and Avail do not.
The Celestia team believes that tokens are necessary for security, and EigenDA will eventually require tokens because it is impossible to cut the availability of on-chain data. They believe that in order to ensure that nodes are honest, data is available and punish malicious nodes, the network must be verifiable through an incentive structure that includes native tokens. Here, Celestias Nick White offers a criticism of EigenDA: Revalidators that retain data will not be slashed unless the source chain is forked - which is highly unlikely, since this is Ethereum.
From a brand perspective, EigenDA is a product that is extremely consistent with Ethereum. The EigenLayer team is building on EIP-4844 and danksharding—EigenDA is being built as “the only ETH-centric data availability layer,” in Sreeram’s words. He explained that by definition, the data availability layer is a modular product, but other DA layers are actually the blockchain itself.
Packaging the DA layer into the blockchain does bring obvious benefits to Rollups running natively on it, mainly in the form of security guarantees. However, Sreeram mentioned that his team’s goal in building EigenDA was to create a product that provides data availability services to the Ethereum ecosystem from first principles—a true “layer” adjacent to the Ethereum ecosystem. He noted that there is no need for a separate consensus here, as Ethereum-based Rollup already relies on the network for ordering and consensus. (Sreeram explained this in a recent episode of Bankless.)
Avail is built with proof-of-validity and DAS to enable a high degree of flexibility and interoperability in the ecosystem. Their architecture lays the foundation for an extensible framework designed to support services across many different platforms. This “unbiased” stance allows for greater interoperability and capital flows, and also appeals to non-Ethereum-centric ecosystems. The ultimate goal here is to get ordered transaction data from all chains and aggregate it into Avail, making them the coordination center for all web3. To launch the network, Avail recently launched a node clash event alongside its incentivized testnet, allowing users to run validators and light clients and participate in network challenges.
Celestias ecosystem consists of RaaS providers, shared orderers, cross-chain infrastructure, etc., covering ecosystems such as Ethereum, Ethereum rollups, Cosmos, and Osmosis.
Each of these design choices, whether technical or marketing, comes with interesting trade-offs. Personally, Im not sure if the data availability category will be a winner-take-all or commoditized market - instead, there may be an oligopoly-type market where projects choose the DA layer that best suits their needs. Depending on the type of protocol, teams can optimize for interoperability, security, or preference for a certain ecosystem or community. If custom use case aggregation explodes as expected, they wont hesitate to integrate the DA layer - and there will be more than one powerful option to choose from.
The technology – and the modular narrative in general – is still relatively new, with Celestia just recently going live and Avail and EigenDA set to hit mainnet in the coming months. However, the technological advances in modularism so far have been remarkable (many of these concepts were just ideas a few years ago!). By fundamentally improving the way we build and use blockchains, the DA layer will undoubtedly become one of the core technologies of this cycle and beyond.
