Research into the effects of a cluster size limited mempool in 2023

Research into the effects of a cluster size limited mempool in 2023

Original Postby sdaftuar

Posted on: April 12, 2024 20:28 UTC

The overview of a new mempool design for Bitcoin and its potential impacts on network transaction handling has been the subject of significant analysis.

A prototype cluster mempool implementation was compared against the current Bitcoin Core design through simulations using data from 2023. This innovative approach recorded every transaction, block, and related data received over the network for extensive periods, allowing for a detailed comparison of how each mempool design would perform under identical historical conditions.

The methodology employed in this study involved enabling datalogging and simulation modes to replay network activity through Bitcoin Core's validation processes. This setup aimed to assess changes in mempool or consensus logic without actual network participation. The simulations were configured to maximize logging detail and mempool capacity to minimize differences unrelated to the mempool designs themselves. Specifically, the cluster mempool prototype was tested with limits set to allow up to 100 transactions or 101 kilovirtual bytes per transaction cluster, unlike the traditional mempool which imposes restrictions based on ancestor and descendant counts.

Results from the simulation indicated a negligible difference in the overall number of transactions accepted between the two implementations, despite variations in rules for transaction acceptance and replacement. Notably, the cluster mempool's unique transaction reprocessing mechanism after blocks are delivered showed that a majority of transactions rejected due to exceeding cluster size limits could eventually be accepted into the mempool. This suggests minimal impact on user experience in terms of transaction relay success rates.

Furthermore, the economic implications of adopting a cluster mempool were analyzed through comparisons of fee differences between the blocks produced in each simulation. Despite minor variations, there was no evidence to suggest that the cluster mempool would significantly affect miners' revenue from fees. The analysis also delved into the Replace-By-Fee (RBF) policy differences, concluding that the cluster mempool's approach to RBF might prevent certain types of transactions that could otherwise degrade the mempool's efficiency, albeit with minimal overall effect on transaction processing.

The investigation into RBF rules highlighted specific scenarios where the cluster mempool's criteria could reject transactions acceptable under the current policy but potentially harmful to mempool efficiency. However, it also acknowledged the possibility of rejecting beneficial transactions due to non-optimal sorting algorithms used for large transaction clusters. This aspect underscores the need for continuous optimization of mempool management strategies to accommodate evolving network behaviors and use cases.

In conclusion, the proposed cluster mempool design, characterized by its novel handling of transaction clusters and refined RBF rules, seems unlikely to adversely affect network operations based on the simulation results. While offering a theoretical improvement in mempool management, the practical implications appear limited but positive, suggesting a potential for enhanced transaction processing without compromising the network's economic dynamics.