helium
Explore using a series of rocksdb checkpoints instead of having a lagging ledger
It's possible that moving to a series of checkpoint ledgers instead of having one double ledger with a number of intermediate states would be more efficient and safer (since we wouldn't have to recompute the intermediates on startup. it would definitely lead to less complicated ledger code.
It's also possible that this would use a ton more memory and disk, since each one would be a separate rocks instance, or would hurt performance in some unanticipated way.
It's fairly clear that this approach works! However, we will need to re-evaluate our cache invalidation strategy, since currently we delete all of these checkpoints and rebuild them from the lagging ledger, which would need to change.
👋🏻 Good day, mate. I make the following utilities based on Solidity & Rust with powerful frontend demonstration. 💻Dapps(Swap, Staking, Referral, Farm, Pool, Presale, Reflection, DAO ...) 💻NFT marketplaces (Opensea, Rarible, Metaplex) 💻Backend software Leave msgs if you are in the demand of me.
