Ethereum’s average block size is up 31% year-to-date, a side effect of an increase in the per block gas limit and demand to transact on-chain last spring.
- According to YCharts, the average Ethereum block comes in just shy of 60 KB compared to around 40 KB at the beginning of the year.
- An increase in the block size accelerates storage needs for Ethereum nodes, with the mainchain up from 610 GB to 875 GB as of July 19.
Ethereum’s block size is larger because of a campaign to increase the gas limit – which essentially changes how many transactions can fit in a block – by miners and developers who wanted to process more transactions on-chain given historic demand.
Want more mining insights like this?
Vitalik Buterin also gave a thumbs up to the idea in April after the Berlin hard fork, which changed gas costs for certain smart contract computations and therefore made a change in the total gas limit tenable. On-chain data shows congestion eased immediately afterwards, until May’s price collapse which found every person looking for a wider door.
Increasing the gas limit per block hurts node operators, however, who must store all or parts of the transaction history in order to validate the chain.
Luckily, data shows little correlation between the block size increase, blockchain data size and willingness to run an Ethereum node. Ether Nodes data also shows a majority of Ethereum nodes run in the US and Western Europe, likely among those with access to cheaper computing components.
That does not mean Ethereum developers are not keenly aware of the need for node solutions given the only ever increasing demand to transact on that network. One solution, a new client named Erigon, has already been deployed for fast syncs of the full network in under 3 days.
Still, the relegation of node size to miners remains one of Etheruem’s more decisive design decisions. Eth 1.0 node runners will continue to bear the brunt of costs if Ethereum 2.0 continues to be delayed, and even after The Merge, a solution for increasing the general node count could be top of mind.