Thoughts on Data Availability (DA)

In this article, you can find information about the roadmaps of the Celestia, Avail, and EigenDA projects. At the end of the article, you can also find information that will enable you to focus on the correct sectors for 2025 and select the right altcoins.

Why is DA so important?

Let's start with Ethereum. Ethereum can currently provide cheap, fast, and secure services to 1 million users, but its goal is to reach 1 billion users.

As the number of users increases, transaction fees soar, necessitating Ethereum's scalability.
L2 projects form the basis for scaling Ethereum. They pull users onto their platforms, reducing Ethereum's traffic and processing transactions (execution).

They batch/process transaction data and send it to Ethereum. This makes transactions cheaper and faster. However, execution alone is not sufficient. The DA side also needs to scale for transactions to become significantly cheaper.

Who uses DA and how?

There are some differences between optimistic rollups (Arbitrum, Optimism), ZK rollups (zkSync, Scroll, Polygon zkEVM), and validiums (Immutable X, Manta: also known as celestium) for DA.

ZK rollups send ZK transaction proofs to Ethereum along with the data. Even if these rollups don't send the data to Ethereum or Ethereum doesn't provide DA, these proofs can be used to revive and make L2s operational again from the last rejected/problematic block.

zkSync has a special case: it sends "state delta" (a smaller piece of information called state change) to Ethereum instead of transaction data, thereby slightly reducing transaction fees. However, if there's an issue with DA, zkSync needs to be brought back up from genesis, the first block.

Optimistic rollups, lacking transaction proofs, cannot be saved without DA, so they rely on Ethereum. The biggest expense for these rollups is DA, for which they pay Ethereum generously. They are eagerly awaiting the development of EIP-4844 (March 13, 2024).
With EIP-4844, particularly DeFi projects with significant capital flows can increase user numbers significantly with cheap, fast transactions, along with the security provided by Ethereum.

Validiums keep the data off-chain (not sent to Ethereum) and only send ZK proofs, making transaction fees reasonable.

Why don't they send the data?

They don't need to stay in frequent communication with Ethereum (DA is not often needed in areas like GameFi/SoFi) or structures like banks/companies may prefer to keep the data private.

How does DA work?

An example:
Suppose there's an FB-GS match (you can think of this match as a transaction on L2). The spectators in the stands (validators) record the match.

Fans outside the field or at home wonder about the score, whether the referee is officiating correctly, whether their bets are successful, etc.

If people are unaware of the match records, nobody believes the match happened, some say FB won, some say GS won, nobody knows whose bet succeeded/failed, chaos ensues.
The match must be broadcast for people to be informed. If the spectators/validators broadcast the match, everyone becomes aware, and people feel at ease (the real-life equivalent of this example: when you check your Web3 wallet, you see your tokens in the correct amounts).

Ethereum's DA is accessible, transparent, decentralized (many people in the stands record the match and share the footage).

In a way, Ethereum's image is very clear, leaving no room for error, hence we can say it's expensive.

Celestia and Avail offer different approaches, let's look at them now.

Celestia, Avail, and EigenDA: How do they offer solutions?

Celestia and Avail use "erasure coding" and "DA sampling" for DA. Erasure coding: think of it as dividing the match into 90 or more pieces.

This way, even light nodes/people watching the match at home become part of the records. Light nodes don't store all match data; DA sampling from within the match: taking samples (1-minute recordings) and keeping them.

As light nodes, we can also temporarily store transaction/match records through our phones. If a lot of people record match data, decentralization & cost-effectiveness are achieved.
When this data is needed or if there's an attack on the blockchain (i.e., an attempt to overwrite the recorded match), light nodes object, preventing this and ensuring blockchain security through DA sampling.

Additional information about Celestia: 0.4% data samples (DAS) are taken in the blockchain, providing 99% data security. In case of objection to block data (similar to fraud-proof logic in optimistic rollups), validators can reject blocks containing faulty/incorrect data. Celestia operates with 100 validators for blockchain security and consensus/block validation.

Additional information about Avail: Similar to Celestia, high data security is achieved with very few data samples using DAS (similar to fans recording 1-minute segments). Developed with Polkadot SDK. It's set to move to the mainnet (estimated date 26.02.2024). They aim to have 1000 validators on the chain, with a goal to increase to 10,000. They're moving towards ZK proofs with erasure coding & KZG commitment. They embed these ZK proofs into block headers to ensure data integrity.

If you've noticed, Celestia leans more towards optimistic rollups, while Avail is akin to ZK rollups.

EigenDA: A project developed on EigenLayer, focusing on providing DA for Ethereum rollups only. It's not an independent blockchain like Celestia or Avail; it's an EigenLayer service. In short, you can think of it as an Ethereum extension.

How does it work? L2 rollups will send transaction data to EigenDA. EigenDA will function similarly to Avail (KZG + erasure coding). It will send this data to EigenDA operators. After obtaining signatures from operators, they will temporarily publish/make this data accessible. As Ethereum is used for data security, there won't be a token specific to EigenDA.

Arbitrum and Optimism will continue to use Ethereum for DA. For Arbitrum Orbit (L3) projects, testing is underway with EigenDA.

EigenDA is expected to move to the mainnet by June 2024. Similarly, projects developed with OP Stack can also use EigenDA (for example, Mantle).

The development of EigenDA's mainnet is particularly positive for Arbitrum, which has a strong ecosystem. The number of Orbit projects will also increase.
Question/Answer:

Why does Manta Pacific use Celestia for DA when Ethereum is available?

Ethereum DA is expensive. Celestia is cheaper. Ethereum is only used to send ZK proofs/settlement. Teams developing applications find this structure beneficial.

If you were a developer, would you choose Arbitrum or Manta?

If you're developing a DeFi-focused application, you need more security (Arbitrum), but if you're developing a game, you need to attract more users (Manta). Of course, there are many parameters; you might also choose Arbitrum L3 for a game or go for Scroll for DeFi, etc., but this is a rough assessment.

Why are Celestia & Avail cheaper?

They use erasure coding and DA sampling. In short, they divide the data into pieces and only take samples from within. This makes them cheaper and appealing for validium & L3 & rollup projects.

Is data availability the same as data storage?

DA provides temporary access to transaction records. For example, with the Ethereum EIP-4844 improvement, approximately 30 days of DA is provided for rollups.
For permanent storage of data, full node validators & decentralized storage projects are used (e.g., Filecoin, Arweave).

What can we expect in 2025 and beyond?

Micro rollups will come. There will be hundreds or even thousands of rollups & rollapps (some will have only one application). The number of L3 projects on optimistic or ZK rollups will increase.

When L3, L4, L5 projects come to Ethereum and rollups, these projects will use solutions like Celestia, Avail, and EigenDA for DA.

Improvements in DA sampling may be made to expedite cross-rollup bridging and make liquidity more efficient - DA sampling for multiple block data could be performed in one go. This could further reduce transaction costs.

The importance of bridges working for rapid cross-rollup & rollapp liquidity transfer will increase. Projects like Hyperlane, Axelar will be crucial.

As the number of modular blockchains increases rapidly, the importance of the settlement layer will increase. Projects like Dymension, Initia, Berachain, Osmosis will be important here.
Expecting people to keep up with inter-bridge & inter-app transitions is difficult. People fear transaction fees & clicking on incorrect links, interfaces aren't user-friendly, etc.

Therefore, projects working on account abstraction will be important (such as managing wallets on the same page, using face ID for wallet security, etc.). Blockchains that bring applications to phones will have a significant advantage.