The State of Eth2, Jan 2021 [ENG/普通话]


At the start of December, the Ethereum community bootstrapped the new beacon chain proof-of-stake consensus mechanism – the foundation of Ethereum’s long-term and sustainable home. The launch was a resounding success, and the operation of the beacon chain has proved, thus far, stable and robust.

As I write these words, the beacon chain is live with more than 2.5M ETH securing its consensus. So this is the “state of eth2” today. But what’s next? What’s on the horizon, and where are we going?

This post aims to provide high level context on the state of Ethereum’s eth2 upgrade – an overhaul of Ethereum’s core consensus to provide a more sustainable, secure, and scalable home for Ethereum and its community.

This post does not go deep into technical details, nor does it aim to be comprehensive. If you’re in search of a more granular account of all things eth2, I recommend checking out Ben Edgington’s What’s New in Eth2 and my “eth2 quick update” series on the EF blog.

Disclaimer: this is how I personally see things today. There are many voices and opinions driving the ever growing, ever evolving eth2 research process. This is just a snapshot of a slice of an interpretation. Enjoy!

eth1 + eth2 == Ethereum

The terms “eth1” and “eth2” are misnomers. If I had my way, I’d toss them out forever, but the terms are pretty ubiquitous at this point so let’s at least discuss what they mean.

These designations – eth1 / eth2 – imply a false sequentiality. They imply that eth1 will be deprecated in favor of eth2, that one is being left behind for the other. Instead, the terms actually represent different layers of the stack.

Eth1 is primarily the operation and upgrading of Ethereum’s user-layer – state, transactions, accounts – all the things the end-user considers when interacting with Ethereum. Eth2 on the other hand is a series of upgrades meant to overhaul Ethereum’s core consensus – to move from the energy-hungry, inefficient proof-of-work to a more sustainable, scalable proof-of-stake. And what do we primarily want to come to consensus on with this upgraded consensus mechanism? The eth1 user-layer!

This relative isolation of concerns (user-layer vs core consensus) has allowed for Ethereum (eth1) to keep chugging along while simultaneously supporting a major overhaul of Ethereum’s core consensus (eth2) in a relatively de-risked fashion. And, in the future, this isolation of concerns might support more specialization across the stack and software components. The heroes (eth1 client devs) that keep Ethereum alive and well today can continue to focus on user-layer optimizations and stability, while the eth2 engineering teams can do what they do best – build sophisticated proof-of-stake consensus.

After the merge, an Ethereum client will use the best from eth1 and eth2 in a modular fashion. For a high level discussion on what this might look like in practice, check out my ethresearch post on the eth1+eth2 Client Relationship (also check out my ‘eth1 + eth2 = Ethereum’ talk at ETHOnline where I belabor this point).

eth2 == consensus

When we talk about “eth2”, we mean software and protocol upgrades of Ethereum’s core consensus mechanism. Foundationally, eth2 is the swap of Ethereum’s consensus from proof-of-work to proof-of-stake and the addition of a sharded data-layer for scalability gains. If proof-of-work is the rocket ship blasting Ethereum out of Earth’s atmosphere, then eth2’s proof-of-stake beacon chain is the permanent space habitat where the rocket will eventually dock for a sustainable and spacious home.

If you take a look at the sophistication that is an eth2 client today, it is entirely in the management of consensus – finality, fork choice rule, validators, and incentives.

eth1 == user_layer

When we talk about “eth1”, we mean the Ethereum proof-of-work chain and all of the existing applications and tools deployed on mainnet. When we discuss upgrades to eth1 and bigger “eth1x” items, we are referencing things almost entirely in the “user-layer”. That is – state, contracts, execution, transactions, etc. If consensus is the brain, the user layer is the body; almost all upgrades to Ethereum since it’s inception have been directed at the body, and the (relatively small) brain of proof-of-work has remained stable.

If you take a look at the sophistication that is an eth1 client today (i.e. geth, nethermind, besu, openethereum, turbo-geth), the lions share of the optimization goes into the management of this user-layer – efficiently accessing/writing the large user state, optimizing execution, safely managing the transaction mempool, etc.

In a nutshell, the Ethereum of tomorrow replaces the brain of Ethereum (with a sustainable and secure proof-of-stake) while keeping the body we know intact. This translates into uninterrupted contracts and applications with stable APIs and tools (see Mikhail’s demo of using an unmodified metamask to interact with an eth1+eth2 merge testnet).

Consensus, but on what?

A common misunderstanding of the launch of the beacon chain is that “it doesn’t do anything”, and my response is “that’s exactly the point!”.

The beacon chain, at it’s core, is a proof-of-stake consensus mechanism secured by validators that have staked ETH. As of today, the only thing that this consensus mechanism comes to consensus on is itself. It does not, yet, come to consensus on anything that end users care about – the user-layer (e.g. transactions, user accounts, smart contracts, etc). This is by design.

The Ethereum network secures hundreds of billions (€/$/Ɖ) in assets and user activity, and thus the path to upgrade its core consensus (although radical in scope) must be conservative and iterative in rollout.

In sum:

Step one [COMPLETED] – come to consensus in a decentralized and secure fashion.
Step two [IN PROGRESS] – come to consensus on items valuable to the Ethereum ecosystem: (a) the Ethereum network/state of today (i.e. eth1) and (b) the sharded data of tomorrow.

Consensus on Ethereum (eth1)

First and foremost, eth2 is for Ethereum. The beacon chain was bootstrapped and is run by the community and ultimately is for the community and its existing applications, contracts, and tools.

The hotswap of Ethereum’s consensus from proof-of-work to the beacon chain’s proof-of-stake will provide a more secure and sustainable home for both today and tomorrow’s decentralized applications. By swapping the consensus mechanism while holding the rest of the protocol (the user-layer) stable, Ethereum will provide an uninterrupted and continuous experience for existing users and applications. Before the swap, applications will continue to be secured and processed by proof-of-work, while immediately after the swap they will be secured and processed by the beacon chain. Existing APIs and tools will largely be unchanged and in general applications will just work.

This work is in progress with a merged geth+teku (eth1+eth2) client running private testnets today. We’re looking forward to releasing a more public playground soon along with some early specifications so that more clients (both eth1 and eth2) can get in on the action.

The safety and stability of the beacon chain

Through 2021, the beacon chain will continue to be battle-hardened, evaluated, and refined for the eventual merge of eth1 into eth2.

One of the necessary criteria for the merge is time. The beacon chain needs to run in production for a sufficient stretch of time to demonstrate its stability and robustness. Beyond that time-stability requirement, I expect detailed network/load analysis, additional testing/fuzzing, and extensive merge testnets.

Sharded data

In addition to a secure and sustainable home, Ethereum also needs a scalable home.

It is abundantly clear that Ethereum L1 today will not, alone, provide the throughput needed to support the global demand for decentralized applications. Although Ethereum’s current L1 coupled with L2 scalability techniques (rollups, channels, etc) will help massively in the next 12 months, even then Ethereum will continue to see demand outstrip supply as global adoption continues.

To complement the exciting L2 rollup ecosystem that creates scalability through the use of L1 data, eth2 aims to come to consensus on a scalable, sharded data layer.

That said, coming to consensus on sharded data is much simpler than coming to consensus on sharded state, both in terms of core consensus complexity as well as the complexity exposed to the end user. At the same time, this scalable L1 data multiplies the scalability gains of L2 rollup adoption.

There is optimism that the merge plus sharded data will provide functional escape velocity and ultimately get us to where we want to be. That is not to say that a sharded user-layer (user state and transactions across multiple shards) is entirely out of the question, just that today it is currently on the back-burner as we bring simpler yet powerful upgrades to Ethereum.

Check out Vitaliks post – A Rollup Centric Roadmap – or his talk – Scaling Ethereum in 2020 and Beyond for more behind the reasoning here.

Upgrades along the way

A first set of upgrades to the beacon chain is actively being spec’d and discussed. Some of the changes proposed are iterative improvements to validator incentives and state management, while others bring new, but relatively modest, features to the beacon chain – e.g. native light client support.

Much of the conversation around these iterative upgrades is taking place in the eth2 specs repo as well as in the Eth R&D discord. I expect client teams and other contributors to share more about these as they solidify in the coming month.

This first upgrade to the beacon chain will also serve as something of a “warm-up” for both eth2 client developers, as well as for the community of stakers. With a more ambitious upgrades of the merge and sharding in sight, starting with a smaller upgrade is prudent.

The state of eth2 clients

In general, we are in a great place with eth2 clients. We have four production mainnet clients (Nimbus, Teku, Lighthouse, and Prysm), each with a share of validator usage. Additionally, Lodestar is currently running mainnet in more of an alpha state while continuing to provide excellent JS tools and libraries to the growing eth2 developer ecosystem.

There is continued work on performance and stability across the board, and there are still gains to be had in terms of client diversity. But all in all, we’re seeing a rich ecosystem of both clients and stakers.

Client diversity

As far as we can tell, client diversity is not optimal today. According to some estimates, Prysm nodes account for at least 50% of mainnet nodes, and although this does not map 1:1 to the amount of stake being secured by each client, it is likely directionally representative.

With four production mainnet clients, the ideal distribution is closer to equal weight (25%) across all clients, and certainly less than 50% for any one client. Given the state of eth2 clients, this is achievable, but it will take the concerted effort of community members and institutional stakers to seek out and adopt the clients with lower shares of the network. As someone who has operated all four clients in the past, I can attest – they are all stable and performant, and each has an enthusiastic and welcoming team dedicated to making life as easy as possible for you.

Client development in 2021

Client teams will spend much of 2021 improving security and stability of mainnet clients while at the same time pushing toward production implementations of the various upgrades discussed above. In early 2021, much of this will be R&D as specs are vetted and refined, but in the latter half of the year, we’ll begin to see production testnets of merge and/or sharding upgrades.

I expect client teams will keep you updated much better than I can. Join their discord servers, follow their repos, and keep an eye on their blogs.

Additional context

Stateless Ethereum

As discussed above, “eth2” broadly represents a series of upgrades to Ethereum’s core consensus mechanism, whereas “eth1x” represents R&D and upgrades to Ethereum’s user-layer. Stateless Ethereum (and much of eth1x) is focused on making the L1 state more sustainable, while opening up a spectrum of node-types that ultimately provides users with more choice on how to interact with the network.

Due to the separation of concerns (consensus vs. user-layer) this R&D effort is being designed and built in parallel to the eth2 upgrades. Although there are some points at which we must ensure the designs are cohesive, statelessness can be shipped either before or after the eth1+eth2 merge. The sooner the components of statelessness ship, the better, but eth2 is not dependent on the stateless timeline.

There are currently monthly stateless calls and an active R&D effort. I expect to see some solid movement on some of these iterative upgrades in 2021!

EIP 1559

EIP 1559, a highly anticipated upgrade of Ethereum’s fee market, is also sufficiently independent of eth2 and can technically happen either before or after the eth1+eth2 merge. That said, R&D on this item has picked up steam in the past 12 months, and we optimistically will see 1559 fee mechanics on mainnet in 2021.

Something of note, though, is that 1559 style fee mechanics will be native to eth2’s sharded data layer. This will allow for a better UX to Ethereum’s data consumers when attempting to chuck data into the Ethereum network, most tangibly aiding rollups in confidently checking their block data into the beacon chain.

For more on the 1559 upgrade in eth1, check out Tim Beiko’s great 1559 Updates.

Community and tooling

The community and tooling around eth2, although still early, is currently thriving and rapidly growing. There are many community organized initiatives – EthStaker, Ethereum Due Diligence Committee, Ethereum Studymaster, Secret Shared Validators, and more – all actively enhancing the validator experience (and new open source projects and contributions to tooling each week)!

The EF ran a very successful Medalla Data Challenge and is in the midst of evaluating a ton of exciting proposals from the Staking Community Grants round. Keep your eyes peeled for more exciting eth2 initiatives this year!

With the launch of the beacon chain, there is seemingly endless green field to get involved and make an impact. Remember, we’re building this together. If you feel something is missing, or a project could use some help, please don’t hesitate to get involved! Contributions, both small and large, can make a huge impact.

Wrapping it up

All in all, there is a ton of work to do in 2021, but what’s new?

The successful launch of the beacon chain was a huge milestone for engineers, stakers, and the community at large. We now have the foundation for Ethereum’s sustainable future. Although there are many moving pieces and innumerable winding threads, across the board the Ethereum community has shown its unbelievable resilience and ability to deliver.

Here’s to a great 2021

12 月初,以太坊社区启动了新的信标链权益证明共识机制,这是以太坊长期可持续发展的基础。 信标链的启动取得了巨大的成功,到目前为止,信标链的鲁棒性良好,运行是稳定的。

在我写这篇文章时,信标链上已经有超过 250 万个 ETH , 在不断巩固共识。所以,这就是当下“ eth2 的状态”。 但是接下呢? 会发生什么,我们又要去哪里?

这篇文章旨在从一个较高的视野向读者描述以太坊 2.0 的景况——对以太坊共识机制进行一个全面的审视,以便其能继续为以太坊及其社区提供一个更加可持续、安全且具有扩展性的家园。

这篇文章不会深入技术细节,也不旨在全面。 如果您想更详尽地了解 eth2 的其他方面,建议您在以太坊基金会博客上查看 Ben Edgington 写的“eth2 的新功能”以及我写的“ eth2 快速更新” 系列。

免责声明:这些只是我目前个人的观点。还有很多其他的声音和观点都在推动着 eth2 不断发展与进化。这篇文章您可以看作是对动态发展进行阐释的一张掠影快照。

eth1 + eth2 == Ethereum



eth1 主要是对以太坊用户层(状态,交易,账户)的操作和更新,以及用户最终在与以太坊进行交互时会考虑的其他所有内容。另一方面,eth2 是一系列升级,旨在彻底改变以太坊的核心共识——即从能耗巨大,低效的工作量证明过渡到更可持续,可扩展的权益证明。而我们通过这种新的共识机制,最想就什么方面达成共识呢?答案是 eth1 的用户层呀!

这种两种相对独立的关注点(用户层与核心共识)使得在以太坊(eth1)有序地向前推进着的同时,还能支持以太坊核心共识(eth2)以相对低风险的方式进行重大改革。而且,在未来,这种关注点的相对独立性也可能使得整个以太坊堆栈的不同组件变得更专业化。如今,那些使得以太坊保持活力与良好状态的英雄们(eth1 客户开发人员)可以继续专注于优化用户层和增强稳定性,而 eth2 工程团队可以尽其所能去建立完善的权益证明共识。

eth1 与 eth2 合并后,以太坊客户端会以模块化的方式利用 eth1 和 eth2 各自最优的内容。关于合并后在实践中具体会是什么样子,我们曾有一场高质量的探讨,具体内容可以查阅我在“eth1+eth2 客户端关系”话题下发表的以太坊研究文章。 (也可以查看我在 ETHOnline 上“eth1+eth2= Ethereum”的演讲,我曾就此发表过看法


当我们谈论“eth2”时,我们是在说以太坊的核心共识机制的软件以及协议升级。从根本上讲,eth2 是以太坊共识机制从工作量证明到权益证明的转换,并且引进了分片数据层来提高扩展性。如果说工作量证明是一艘火箭飞船将以太坊带出地球大气层,那么使用权益证明的 eth2 信标链则是永恒的太空栖息地,火箭最终将停靠在那里,形成一个可持续的广阔家园。

如果您看一下当今 eth2 客户端的复杂程度,会发现它完全取决于对共识的管理,例如确定性,分叉选择规则,验证者和激励措施。


当我们谈论“eth1”时,我们是在说以太坊的工作量证明链以及所有在主网上开发的应用程序和工具。当我们讨论 eth1 以及更大的“eth1x”项目升级时,我们谈到的几乎都是“用户层”里的事情。比如状态、合约、执行和交易等。如果将“共识”比喻成大脑,那么“用户层”就是身体; 自以太坊诞生以来,几乎所有升级都针对“身体”,而代表工作量证明的 “大脑” (相对较小)则一直保持原状没有什么改变。

如果您看一下当今 eth1 客户端的复杂程度(比如,geth , nethermind, besu, openethereum, turbo-geth会发现优化的绝大部份工作都指向了用户层的管理,比如——高效地访问/编写大量的用户状态,优化执行以及安全地管理交易内存池等等。

简而言之,未来的以太坊会替换掉现在的“大脑”(借助可持续且安全的权益证明),同时保持我们熟悉的“身体”的完整。 这些将通过稳定的 API 和其他工具转化为不受干扰的智能合约和应用程序。(请参阅 Mikhail 带来的演示——使用未经修改的 metamask 与 eth1 + eth2 合并测试网进行交互 。



信标链的核心是一个权益证明共识机制,该机制由质押了 ETH 的验证者来共同维护。截止到今天,这种共识机制唯一达成的共识就是它本身。 对于最终用户关心的其他事情——例如用户层(交易,用户帐户,智能合约等),它尚未达成共识。 这是设计使然。

以太坊网络容纳了数千亿资产(€/ $ /Ɖ)和千万用户的活动的,因此升级其核心的共识方式(尽管所涉及的范围会很广)必须是保守的、逐步进行迭代。

第二部【进行中】—对以太坊生态系统有价值的事项达成共识:(1)目前以太坊网络/状态(即 eth1)和(2)未来的分片数据


首要的一点,eth2 是为以太坊服务的。信标链是通过 bootstrap 的方式启动的,靠社区运行,而其最终也将服务于社区及其现有的应用程序、合约以及其他工具。

从工作量证明到信标链的权益证明,以太坊共识的热交换将为当今和未来的去中心化应用提供更安全、可持续的成长土壤。通过改变共识机制,同时保持其余协议(用户层)的稳定,以太坊将为现有用户和应用程序提供不受干扰的持续性体验。在转换之前,现有的应用程序仍然受工作量证明程序的保护,而在转换之后,将立即由信标链来处理与保护这些应用程序。现有的 API 和其他工具在很大程度上将保持不变,总的来说,现有应用程序会正常工作。

如今,通过使用合并的 geth+teku(eth1+eth2)客户端来运行私密的测试网,这项工作正在紧锣密鼓地被推进着。我们期待尽快发布更多公开场所以及一些早期规范,以便更多客户端(eth1 和 eth2)参与进来。


2021 年,信标链将继续锤炼、评估和完善,最终实现 eth1 与 eth2 的并轨。并轨的一个必要条件就是时间。信标链需要在实践中运行足够长的时间,以证明其运行的稳定性和鲁棒性。 除了时间稳定性要求之外,我还希望进行详细的网络/负载分析,其他测试/模糊测试以及广泛的合并测试网。



显然,全球对去中心化应用程序需要所产生的需求量,单独地依靠如今的以太坊 L1 难以满足。尽管以太坊当前的 L1 配合 L2 扩展性技术(rollups,channels 等)会在未来 12 个月内起到极大的帮助作用,但随着全球采用率的不断提高,在以太坊需求超过供给的现象仍将持续。

L2 rollup 生态系统利用 L1 的数据获得了扩展性,为了完善这个令人激动的生态系统,eth2 旨在就可扩展的、分片的数据层达成共识。

就是说,对于核心共识的复杂性以及最终用户所面临的复杂性而言,就分片数据达成共识要比就分片状态达成共识要简单得多。同时,可扩展的 L1 数据使得采用 L2 rollup 获得了更多的可扩展空间。

存在一种乐观的看法认为,eth1 和 eth2 并轨加上分片数据会让我们加速告别旧世界并最终使我们到达想去的地方。 这并不是说用户层分片(跨多个分片的用户状态和交易)是完全不可能实现的,只是说,在今天它有一些落后了,因为我们为以太坊带来了更简单但功能强大的升级。

如果想了解背后更多的信息,请查阅 Vitalik 的文章— Rollup 的中心路线图(A Rollup Centric Roadmap ),或者他的演讲—2020 年及以后的以太坊扩展事宜(Scaling Ethereum in 2020 and Beyond )。



关于这些迭代升级的许多讨论都发生在论坛 eth2 specs repo以及Eth R&D 聊天室中 。我期待客户端团队和其他贡献者可以在接下来一个月的工作中分享更多有关这些方面的信息。

信标链的首次升级可以看作为 eth2 客户端开发者与参与社区质押的人举办的一次“热身”活动。因为合并和分片的更宏大升级都是从谨慎的小升级开始的。

eth2 客户端状态

总的来说,eth2 的客户端状况良好。我们有 4 个主网客户端(Nimbus, Teku ,Lighthouse 和 Prysm ),每个客户端都可以运行验证器。此外,Lodestar 目前仍是以内测版在运行 Mainnet,但同时也在为不断发展的 eth2 开发者生态系统提供着出色的 JS 工具和库。提高全面性能和稳定性工作仍在继续,并且在客户端多样性方面仍有期待。总而言之,我们将看到客户端与质押者的丰富生态圈。


据我们所知,目前的客户端多样性还不是最优的情况。根据一些估计,Prysm 节点至少占主网节点的 50%,尽管这是 1:1 对应着客户端确认的质押数量,但它仍是一个有所偏颇和具有指向性的代表。

对于主网的四个客户端,理想的分配权重接近相等(25%),对于任何一个客户而言,权重肯定不能超过 50%。考虑到 eth2 客户端的状态,这是可以实现的,但需要社区成员和机构质押者共同努力,寻找并采用那些所占权重较低的客户端。作为用过这 4 个客户端的人,我可以证明它们都很稳定,表现都很出色,而且每个客户端背后都有一个热情友好的团队,他们期待着与你一起合作。

2021 客户端发展

客户端团队利用 2021 年的大部分时间来改善主网客户端的安全性和稳定性,同时推动上述各种升级的落地实现。在 2021 年初,随着规范的通过审核和逐步完善,以上的大部分工作将进入研发阶段,但是在下半年,我们将开始看到并轨和(或)分片升级的生产测试网络。



如上所述,“ eth2”广义上代表对以太坊核心共识机制的一系列升级,而“ eth1x”则代表对以太坊用户层的研发和升级。无状态以太坊(以及大部分 eth1x)致力于使 L1 状态更具可持续性,同时开放一系列节点类型,从而为用户与网络进行交互提供更多的选择。

因为关注点(共识与用户层)的相对独立,该研发工作被设计为与 eth2 升级并列进行。尽管在某些方面我们必须确这样的设计是连续顺滑的,但是“无状态”在 eth1 + eth2 合并之前或之后都可以进行。虽然无状态组件越早交付越好,但是 eth2 并不依赖于无状态的完成进度。

当前,每个月都有关于“无状态”的消息发布以及其他积极的研发工作在进行。 我希望 2021 年在这些迭代升级中的能看到稳步的推进!

以太坊改进提案 1559

EIP 1559 是以太坊备受关注的改进提案,有关于以太坊的收费市场。它也完全独立于 eth2,并且从技术上来说,可以在 eth1 + eth2 合并之前或之后进行都可以。过去 12 个月,该产品的研发工作日渐活跃,我们乐观地预计,到 2021 年,主网上将出现 1559 提案的收费机制。

不过,值得注意的是, 1559 提案中的收费机制和 eth2 分片数据层天然地很契合。当以太坊的数据使用者尝试将数据插入以太坊网络时(最切实地帮助 rollups 安全地将其区块数据登记到信标链中),会给他们带来更好的用户体验。

有关更多 1559 针对 eth1 的升级提案的信息,请查阅 Tim Beiko’s 写的“1559 更新”


尽管 eth2 的社区和工具仍然处于早期阶段,但是它们正在蓬蓬迅速的发展着。有很多社区发起的倡议活动——例如 EthStaker , Ethereum Due Diligence Committee, Ethereum Studymaster , Secret Shared Validators 等等,这些都显著地优化了验证者的体验(对新的开源项目以及工具开发的每周进展亦有裨益)。

以太坊基金举办了一场非常成功的 Medalla 数据挑战赛 ,并且他们正在评估来自社区质押捐赠活动(Staking Community Grants )中的大量激动人心的提案。请睁大眼睛,敬请期待今年更多令人兴奋的 eth2 活动吧!



总之,2021 还有大量的工作需要完成,但是有无新鲜事呢?信标链的成功启动对于众工程师、质押者以及整个社区而言是一个巨大的里程碑。如今,我们为以太坊的可持续未来奠定了基础。虽然还有一些不太坚固的垣壁和数不清理还乱的纠缠线索,但是以太坊社区已经全面展现了其难以置信复原力及交付力。
祝福伟大的 2021!

Source link


Please enter your comment!
Please enter your name here