As of August 2022, a 1TB SSD is getting tight to run an Ethereum node in. The recommendation now is for 2TB of space. I wanted to see what it takes to still fit into 1TB.
EIP-4844 will increase space need by another 200 GiB or so, which may make 1TB setups plain not feasible. This is a 2023 discussion however.
- Geth with offline prune every 2-3 months ... this would be a stop-gap, as offline prune takes the node down, including any validators on it
- Future Geth without prune ... this might work well.
Any use of Geth on a staking node comes with risk for the node operator, as long as Geth still has a supermajority (greater 2/3rd) of the network.
-
Nethermind with online prune ... this takes some careful setup, and can work. Run Nethermind with memory prune, when disk space is low, stop it, run it with full prune, manually kick off an online prune, when prune is done, stop it, and GOTO 1.
-
Future Nethermind without prune ... timeline unknown
-
Future Besu with storage optimizations ... even with Bonsai tries it grows at ~ 8 GiB/week, and it cannot be pruned
-
Erigon without prune, on ZFS w/ lz4 and 16k recordsize... the commit cycle is too slow for this to be useful
Keep an eye on EIP-4444, which will drastically reduce the need for state storage.