Been following the Ethereum node infrastructure discussions lately, and there's actually a pretty significant issue that doesn't get talked about enough. The storage bloat on full nodes has become a real pain point for operators. Marek Moraczyński from Nethermind raised something interesting recently about how over 80% of a full node's disk space is just sitting there as historical records from before the merge.

What caught my attention is how they're tackling this in version 1.3.1. They've figured out a way to prune those pre-merger historical records using ERA files, and apparently all the major teams in the Ethereum ecosystem have aligned on implementing this around May. If you strip out all the old blocks and receipts and just keep what you actually need to validate the current state, you're looking at nodes running under 200GB instead of the current bloated setup.

This is actually pretty clever engineering. Most people don't realize how resource-intensive running a full node has become. You've got Geth and other clients dealing with the same storage constraints, and this kind of optimization is exactly what the infrastructure layer needs. Kessler's been writing about these node efficiency issues for a while now, and this feels like real progress.

The practical impact? More people can actually run full nodes without needing enterprise-grade hardware. That matters for decentralization. It's one of those unsexy infrastructure improvements that doesn't make headlines like a new token launch, but it's probably more important for Ethereum's long-term health.