Vitalik Buterin suggests implementing ‘partially stateless nodes’ to help scale Ethereum

Ethereum co-founder Vitalik Buterin proposed a new roadmap on Monday to solve issues linked with scaling Ethereum Layer-1 through higher gas limits, including a new concept called "partially stateless nodes."

"The most common criticism of increasing the L1 gas limit, beyond concerns about network safety, is that it makes it harder to run a full node," Buterin wrote in his recent post.

Running a full node is valuable as it offers a "trustless, censorship-resistant and privacy-friendly way" for users in accessing the chain, the Ethereum co-founder added.

To scale the L1 gas limit without sacrificing running full nodes, Buterin proposed short-term priorities, which include implementing EIP-4444 that limits nodes to contain only up to 36 days of historical data, reducing disk space for other participants.

Running a full Ethereum node requires storing the entire blockchain state (~1TB for state, ~500GB for history). EIP-4444 would offload historical data storage, making nodes lighter.

Other short-term proposals by Buterin were building a distributed history data storage solution and adjusting gas pricing to make storage more expensive and execution less expensive.

Buterin's roadmap highlighted "stateless verification" as a medium-term change, which could allow nodes to interact with the blockchain without maintaining Merkle branches, which are used to verify data integrity. This could cut storage needs by roughly 50%, making nodes significantly lighter, Buterin said.

In his latest proposal, the Ethereum co-founder unveiled a new concept called "partially stateless nodes." Buterin said these have the potential to increase the L1 gas limit by 10 to 100 times.

According to Buterin, these nodes verify blocks and the entire chain without storing all the data by utilizing stateless verification or zkEVM . They will be programmed to store a selected subset of data instead of a full set, and are still able to perform requests pertaining to data in the selected portion.