Validator Set Concerns and
Proposed Solutions

The Current Situation

After being finalized in April 2023, the Shapella upgrade triggered a major surge in staking demand, prompting discussions within the Ethereum developer community about how to best move forward. Prior to the update to the original staking mechanism, users who chose to stake were doing so without an exit timeline, converting liquid ether into illiquid ether. Since the upgrade, users have been able to enter and exit the staking system freely, with a known time frame. With the lowered risk from increased liquidity, the active validator count has risen by 74%.

This rapidly increasing active validator count is raising concerns surrounding technical capacity as well as centralization; these concerns are detailed below. In response to these issues, one popular proposal has been officially included in the next Ethereum upgrade, while another is pending further community feedback and testing. The proposals suggest a change in the code that will decrease the number of validators allowed to join the network within a specific time frame (churn limit) and increase the effective balance of validators.

What Is a Validator Node? 

Many use “validator,” “node(s),” and “validator node” interchangeably, but there are important distinctions between the terms: 

A node is the hardware required to run a validator. It lives in the physical world and enables communication through the network for data retrieval and validation.

A validator is made up of software and exists in the virtual world. It is a key component of a “validator node,” a term commonly referenced when talking about a node. When we say “validator,” we are specifically referring to the three different software clients that make up the validator: an execution client, a consensus client, and a validator client.¹ The validator is responsible for executing transactions, upholding a consensus, and validating new transactions and blocks.

A validator node is the combination of all the required clients running on a set of hardware. Because a validator is virtually hosted on the node (hardware), a node with strong hardware could potentially run multiple validators. However, each validator acts as a unique attestor (voter) on the network. This means each validator must send and receive transaction data and block data to maintain their version of the current state of the network (blockchain). These blocks occur every 12 seconds, requiring validators to download the block, verify transactions, and vote/attest all within this time frame.² Although attestation submissions are possible after the 12 seconds pass, as long as they occur within the same epoch (6.4 minutes), the reward payout diminishes proportionally to the time taken to attest. The longer a node takes to submit its vote, the smaller the reward.

Why Is a Large Validator Set a Concern?

The payload being uploaded and downloaded to each validator includes different types of information: transaction execution-, reward-, penalty-, slashing-, entry and exit-, and attestation-information from the network (other validators). Each validator must then re-execute the transactions locally to ensure validity.

Bandwidth and latency are critical in a large validator set network, where each validator must independently download the latest data and verify state change proposals within a small time frame (12 seconds). The larger the block (data), the more computing power needed to process and re-execute the transactions before the next slot.

Centralization Risk

Currently, the daily block size is, on average, 1.6 megabits per block.³ That equals roughly one megabyte per minute. Per Statista, the average fixed broadband upload speed is about 35.2 megabits per second—264 megabytes per minute.⁴ Geth, an official Ethereum client, recommends a download speed of at least 25 megabits per second, or 187.5 megabytes per minute and an unlimited data cap.⁵ Using the average household bandwidth and the recommended connection speed, roughly 71% of the average household’s bandwidth would be required for each validator on their network.

It is already difficult for the average household to run a validator node because of the required bandwidth. According to a 2023 study, this may be the leading reason why it is estimated that 75% of nodes were hosted by large entities such as AWS, Google, and OVH versus 25% hosted by solo stakers.⁶  

Each new validator adds an additional connection to the network, increasing the overall bandwidth required to maintain consensus. The potential concern is that as the bandwidth requirements grow, the validators that are unable to keep pace will drop from the network—those that drop are more likely to be the self-hosted nodes. If the average household struggles to keep up with the network, there is a risk of increased centralization over time, as the only hardware to survive may live within institution-owned data centers.

Alternatively, Ethereum's rising requirements may not necessarily lead to increased centralization if technological improvements keep pace with the growth of the network. For example, if the network requirements and technology advance at the same rate, then it is net neutral. However, if technological improvements outpace Ethereum’s network requirements, then the advancement could instead serve as a decentralizing force rather than a centralizing one.

Technical Challenges for the Future Roadmap

Future roadmap upgrades will become more difficult with a larger validator set. Upgrades such as EIP-4844 (Proto-Danksharding) will require validators to send and receive more data. This will likely cause latency between nodes and the network, which could then lead to increased centralization of the network as smaller budget setups will be the first to drop off, due to higher performance hardware requirements, increased bandwidth requirements, and lower yields.

The introduction of a new transaction type called “blobs ” will undoubtably increase data transfer sizes. A blob is simply a new type of data input that “attaches” to a block without changing the block size. These blobs can later be pruned from the blockchain after a certain time period. The current target of blobs is three per block, but the maximum will be set at six. Each blob equals roughly 0.125 megabytes.⁷

As Ethereum continues to scale and pursue new features, the amount of data transferred on the network may also increase.

Proposal: Limit Validator Set Growth

One of the accepted proposals sets a cap on new staking nodes and aims to slow the quickly expanding validator set. The proposal (EIP-7514) limits the churn limit to eight validators per epoch or 1,800 validators per day.⁸ 

Due to the exponential behavior of the churn limit, this problem could quickly get out of hand if left unattended. Fortunately, an upgrade (Dencun) was successfully implemented in Q1 2024 and this proposal was included within it.⁹ Validators are still able to join the network, but only at the decreased rate. This means the proposal will only work in the short term and is not intended to solve the actual problem of an increased validator set. However, it will allow more time for further research into the secondary effects of the proposal to increase the effective balance discussed below as well as any necessary tests. 

Proposal: Increase the Max Effective Balance

This proposal is intended to alleviate concerns related to the large validator set, in terms of latency and network size. Looking ahead, increasing the maximum effective balance could reduce the number of validator connections within the network.

The proposal aims to decrease the current validator count by simply allowing a higher effective balance. A validator currently requires 32 ether, and for every additional 32 ether, a user must set up another validator, introducing another hop to the consensus network. This does not mean that a user needs additional hardware; validators are virtual, and many can exist on one “computer.” However, each validator will need to validate new blocks and re-execute transactions before passing the data to another validator through the peer-to-peer spread of transactions (gossip protocol).   

The 32-ether maximum effective balance artificially inflates the validator set size. Large operations are forced to run thousands of validators due to users’ staking demand, resulting in a large and rapidly growing validator set. Increasing the effective balance to 2,048 ether allows large operations not only to reduce their validator footprint but also to simplify their key management.  

Today there are approximately one million validators, meaning there are 34.5 million ether being staked. If Ethereum were to increase its effective balance maximum to 2,048 ether, then the network could see a maximum consolidation of roughly 98%. However, it may not be realistic to think that every operation that can consolidate will do so. This upgrade would purely be “opt-in” with many benefits as well as drawbacks to consider by the operator. Therefore, it would not be mandatory to consolidate. The minimum of 32 ether would not be raised.

With fewer validators would come fewer attestations, and a reduction in bandwidth requirements would follow suit. That translates to quicker signature aggregation, potentially paving the way for faster finality use cases.

A large operation would then be able to run a single validator with 2,048 ether instead of the 64 validators required today. Each validator would be weighted by the amount of ether staked, just as each validator gets one vote today. A validator with 2,048 ether would have the same voting power as 64 validators with 32 ether each. This helps with the gossip protocol by reducing the artificial inflation of validators due to the small effective balance. In the above scenario, there are now 63 fewer validators that need to download and verify the blocks, resulting in lower bandwidth requirements for all nodes.

Conclusion

The large and active Ethereum community has provided substantial feedback regarding these concerns and their respective proposals. The validator set has increased significantly since the Shapella upgrade, and although this growth has recently slowed, there is no telling what the situation may be a year from now. Therefore, the possibility of rapid growth could be a problem due to centralization and bandwidth risks. While one solution has been implemented and another is still being discussed, the former is merely a short-term solution. It is not known whether an increased effective balance would solve the issue over the long term, but a lower churn limit will slow the entry queue for the time being.

It is important to note that the challenge of an increasing validator set has always been viewed as a “good” problem to have because it represents growing adoption and security for the Ethereum network. That said, it is impossible to accurately predict the staking demand in the future. While some may wish for a more natural solution determined by market participants, many in the community do not want to leave it to chance.  

Want to learn more about Ethereum on our platform? Fill out this form to get in touch.