An overview of the cluster mempool proposal

An overview of the cluster mempool proposal

Original Postby sdaftuar

Posted on: January 12, 2024 15:04 UTC

In the realm of cryptocurrency mining and fee rate analysis, a pertinent issue arises when comparing the profitability of different mempool states characterized by their respective fee rates and transaction volumes.

The debate centers on how to equitably evaluate two mempools, referred to as OLD and NEW, which terminate at disparate vbytes. When faced with a scenario where NEW's accumulated fees surpass those in OLD at the point where NEW concludes but fall short by OLD's final transaction, it is suggested that padding NEW to match OLD's size offers a solution. This approach maintains the fee total for NEW constant beyond its original endpoint, rendering the comparison at the larger size moot due to incomparability.

Divergent perspectives exist regarding the "tail feerate," which may be greater than zero, signifying an inexhaustible pool of transactions for miners to include in blocks. Advocates for this concept propose extrapolating the fee rate diagram of NEW based on a chosen tail feerate, allowing for a direct comparison with OLD. However, there are implications when assuming a non-zero tail feerate, particularly when it comes to the possibility of attributing unearned fees within the OLD mempool, which could hinder potential transaction replacements that would otherwise occur with a zero tail feerate.

The current consensus leans towards the notion that a tail feerate of zero is the most defensible approach, given the complexities and theoretical nature of alternative viewpoints. While acknowledging the limitations set by anti-DoS rules requiring an increase in total fees, the discussion remains largely hypothetical. Future research might explore incentive compatibility more deeply, considering the varying significance of fee rates throughout the mempool and potentially leading to software updates that implement refined comparison tests.

On a separate note, implementing a carve-out rule to accommodate certain types of transactions without violating cluster size limitations presents another challenge. The rule aims to exempt qualifying transactions from being counted against a cluster's maximum size or from eviction due to oversize. However, practically any transaction could produce offspring that comply with the carve-out criteria, complicating enforcement. Two primary strategies emerge: evicting existing transactions or expanding the cluster beyond its limits. Both approaches have significant drawbacks, including potential disruption to multiparty contracts, vulnerability to free relay attacks, and increased complexity for validation processes.

The proposal suggests that maintaining strict adherence to predefined cluster sizes is crucial, rather than designing policies that allow exceptions and risk exceeding safe limits. Emphasizing sound engineering principles, the argument advocates for clear and enforceable policies that do not rely on analyzing emergent higher-layer properties.

A plausible alternative involves redefining the policy framework to facilitate transactions conforming to specific topologies, thereby respecting cluster size restrictions. The proposed v3 transaction model encapsulates this philosophy and, if successful in accommodating use cases reliant on the carve-out, would represent a considerable improvement over current methodologies.