delvingbitcoin
An overview of the cluster mempool proposal
Posted on: February 2, 2024 15:00 UTC
The discussion opens with a critique of the current approach to improving mempool quality, suggesting that while there is an improvement, it isn't achieved in the most efficient manner.
A point is made about the untidiness of the process, implying that a more streamlined method should be possible. The inefficiency in question relates to rebroadcasting and revalidating transactions that were unnecessarily evicted from the mempool.
Further exploration into alternative solutions is proposed, such as offloading the computational work onto peers. These peers would be responsible for determining which transactions to evict to accommodate new ones, a task that the original node would simply verify. This suggestion aims at achieving a neater engineering solution, though it immediately raises questions about the feasibility of such a determination by a peer or wallet.
Concluding thoughts question the value of implementing the sibling eviction concept given its complexity and limited scope of application. It's recognized that the idea has inherent flaws, including susceptibility to pinning vectors due to Replace-by-Fee (RBF) mechanisms, and is only relevant within a narrow network topology. This skepticism challenges whether the added complexity justifies the potential benefits, hinting at the need for a reassessment of the strategy for mempool management.