一文搞明白DA 是否会侵蚀以太坊叙事？

作者：Lanli 来源：X，@yorkzhong

1. 模块化区块链的组成部分：

⇒执行层：Rollup/ L2/L3，执行用户/智能合约所需的计算，理解为一台计算机

⇒DA：执行层的所有原始交易确保在这里进行存储，以待后面校验和确认，可以理解为数据库

⇒结算/Settlement：校验/确认/Finality，用户可以在这里发起反驳，需要用到DA里存储的执行层原始数据

⇒还有一层是共识层：这个主要是Layer 1的内容，在这里不赘述。

2. 三个层次的主要技术栈或者代表产品：

• 执行层主要有：

⇒OP Stack

⇒ZK Stack

⇒Cosmos Stack

• 结算/Settlement大约有3种：

⇒Eth layer1

⇒Dymension (与Celestia合作的settlement)

⇒Layer2本身 (例如Arbitrum，可以作为次一级Layer3的settlement)

• DA大约有5种：

⇒ETH blob/ Danksharding

⇒Celestia (独立的去中心化DA)

⇒Near DA

⇒EigenDA/MantleDA (利用 ETH restaking的DA)

⇒中心化DA (Validum模式)

图1是一个Near做的不同DA的比较。这里以太还有一个终极状态的danksharding没在里面，当然也合理，因为以太暂时还没确定danksharding 的时间表（坎昆升级是4844, prototype Danksharding）。

• 从技术角度，以太Danksharding和Celestia是最去中心化的，因为都用到了抽样技术，这样降低了节点的性能要求的同时可以做到大带宽。

• EigenDA次之，也用到了抽样，但是EigenDA是寄生于以太，其节点数是以太的子集。

• 其他的应该没用到抽样，例如NearDA的去中心化程度就相当于Near Protocol的去中心化程度。

3. 模块化区块链的组合方式：

以上大约创造了3 * 3 * 5 = 45种可能的组合，例如：

⇒Arbitrum: OP(执行层技术栈)+ETH L1（结算层）+ETH blob（DA）

⇒Mantle: OP+ETH L1+MantleDA

⇒XAI: OP + Arbitrum L2 + 第三方DA

可谓百花齐放

4. 不同模式的安全性差异

这些不同的组合有什么区别？除了费用和TPS上的区别之外，以太系当然会维护以太的“正统性”，比如Vitalik认为如果不用以太的DA，那么以太没法保证资产能够安全撤出。而他把所有除了以太DA之外的方案统称为"Validum"。（图2）

当然这是一家之言，不过用“短板理论”，比如说你在以太上结算，但是DA却用了另一家（比如near），那么你实质上得到的是Near级别的安全，因为取安全的最小值。

另外一个角度，Celestia的理念之一是轻settlement、重DA、重社会层（哈，就是社区）。

比如说，你是一个celestia上的RU，然后项目方作恶了，这时“社区”可以选择用celestia上的DA做一次硬分叉，完全重构这个链。这也是一个角度吧，不过听起来像是核威慑。

5. 未来的结算层

以下为个人推测，仅作参考：

未来的结算层将十分稀有，大约2-3个主要的结算层。这里涉及几个问题：

⇒资产的跨结算层的跨链仍将会是难题，但是在同一个结算层内的资产跨链相对容易做到安全。

⇒更复杂的是互操作问题。首先互操作是未来必然的一步，web2互操作性非常强，很难想象web3的未来是不能互操作的，那将是极大的退步。而互操作，将要求同一结算层，不同结算层之间因为finality、协议没法协同一致，所以不能进行互操作。Cosmos的IBC也主要强调是IBC体系内的互操作。

⇒另外，结算层的去中心化成本和经济安全要求较高，所以大约只会有2-3个吧。比特网是否能成为结算层之一，还待观察。目前了解到的是比特币的utxo模式尚不够支持做结算，也许未来升级后可以。

6. 未来的DA层

DA层可以已经如雨后春笋出现，核心内容是DA 并没有那么难。最简单的一个单机就可以做DA，最复杂的如Celestia 那样做抽样下的去中心化。

顺便说一下，用抽样模式做的去中心化，节点越多带宽越大（想象一个p2p电影下载网络），这块也是具备网络效应的，所以我认为最终不会有太多这种“去中心化抽样的DA”。

但是其他形式的DA则没有限制，可以有无数个。

在这里我们可以打个比方，DA就是存储，占据成本又很高。如果你不严格的要求以太级别的安保服务，选用哪家DA就是一个成本和安全之间的衡量。

原则是：价值越大的服务，DA应该尽量安全。

因此，未来的DA也许是分散的。即使如此，也许有7-8家主要的DA也许就够用了。

7. 对以太的叙事影响（个人推测）

如果未来的DA层很分散，那岂不是对以太不利？

DA的壁垒确实不是很高，以太的DA有其优势，但也确实存在价格高的问题。但是以太DA仍然可以像苹果那样吃掉最高端、利润最丰厚的部分市场。

以太的未来叙事应该是L2的互操作性，就是说，在以太上的layer2之间，是可以互操作的。这个应该是别的公链很难硬蹭的。

互操作是否能有价值捕获？当然可以，其实ATOM 2.0的 白皮书已经给了一些idea，以太再次借鉴一下也不是什么大问题。

总之先有价值后，捕获总是不难的。

The author's source is the component of modular blockchain. The calculation required by the executive layer to execute the user's intelligent contract is understood as that all the original transactions of the executive layer are guaranteed to be stored here for later verification and confirmation. It can be understood as that the user can launch a rebuttal here and need to use the original data of the executive layer stored in it. There is also a consensus layer. This main content is not repeated here. The main technology stack or representative product executive layer mainly includes There are some kinds of settlement and cooperation itself, for example, which can be used as a secondary level. The centralized mode diagram of independent decentralized utilization is a different comparison. Here, there is an ultimate state of Ethernet that is not in it, which is of course reasonable, because the timetable of Ethernet has not yet been determined. Cancun upgrade is the most decentralized from the technical point of view, because sampling technology is used, which reduces the performance requirements of nodes and can achieve large bandwidth, followed by sampling, but it is Parasitic in Ethernet, the number of nodes is a subset of Ethernet, and others should not be sampled. For example, the degree of decentralization is equivalent to the degree of decentralization. The combination mode of modular blockchain has created about one possible combination. For example, the executive layer, the technology stack settlement layer and the third party can be described as the security differences of different modes. What are the differences between these different combinations? Apart from the difference in cost and cost, the Ethernet system will certainly maintain the legitimacy of the Ethernet. For example, it is considered that if the Ethernet is not used, then the Ethernet. There is no guarantee that the assets can be safely withdrawn, and he refers to all the schemes except the ether as a diagram. Of course, this is a statement of one family, but buckets effect is used. For example, you settle accounts on the ether, but you use another one. For example, then you essentially get the level of security, because one of the ideas from another angle is to emphasize the social layer, that is, the community. For example, if you are on the top, then the project party commits evil. At this time, the community can choose to use it to make a hard fork and completely reconstruct this. This is also an angle, but it sounds like a nuclear deterrent. For personal reference only, the future settlement layer will be very rare. There are about 10 major settlement layers here. It will still be a problem to cross the chain of assets across the settlement layer, but it is relatively easy for assets within the same settlement layer to cross the chain. What is more complicated is interoperability. First of all, interoperability is an inevitable step in the future, and it is very difficult to imagine that the future will not be interoperable. The great retrogression and interoperability will require the interoperability between different settlement layers in the same settlement layer because the protocols cannot be coordinated, and the main emphasis is the interoperability within the system. In addition, the decentralization cost and economic security requirements of the settlement layer are high, so it remains to be seen whether there will only be a bar bit network to become one of the settlement layers. What we know at present is that the bitcoin model is not enough to support settlement, and maybe the future layers can have mushroomed after upgrading. The heart content is not so difficult. The simplest single machine can do the most complicated decentralization under sampling like that. By the way, the more decentralized nodes made by sampling mode, the larger the bandwidth. Imagine a movie download network, which also has network effect, so I don't think there will be too many decentralized sampling in the end, but there are countless other forms. Here we can make an example, that is, the storage occupation cost is high if you are not strict. Which one to choose for Ethernet-level security service is a measure between cost and safety. The principle is that the greater the value, the more secure the service should be, so the future may be scattered. Even so, maybe a major one may be enough to influence the narrative of Ethernet. Personally, I guess if the future layers are scattered, isn't it not a high barrier against Ethernet? Ether has its advantages, but it does have the problem of high price, but Ether can still eat the top profits like Apple. The richest part of the market, the future narrative of Ethernet should be interoperability, that is to say, it is interoperable between Ethernet. This should be whether the interoperability of other public chains is difficult to rub hard. Of course, it can be captured. In fact, the white paper has given some Ethers a reference again, which is not a big problem. In short, it is always not difficult to capture after being valuable. 比特币今日价格行情网_okx交易所app_永续合约_比特币怎么买卖交易_虚拟币交易所平台