Currently our app-side mempool has all transactions, ready to execute or not, inside of the same data structure. This makes mempool maintenance difficult. We have to iterate over all transactions when doing maintenance and there is no clean approach to doing quicker account-based checks.

It'd be nice to change the data structures to something similar to what geth/reth have, including account based transaction storage and a distinction between 'ready to execute' and 'could execute in the future' transactions.

A start to a more efficient mempool would be rewriting our current mempool with similar data structures.