Investigating Mempool Optimizations for Ethereum

Abstract

In this paper, we propose a new scheme OMCEM that optimizes the memory consumption of the Ethereum mempool. Our work will be as follows: at first, we Transform Ethereum mempool to a bloom filter based scheme with some benefits such as it reduces the mempool to a fraction of the current size, and it offers resilience against DDoS attacks, and It also facilitates setting up Ethereum node on low resource devices such as embedded systems and system with low computation resources. Second, we will also perform a simulation on the Ethereum transaction log of over one month and offer a detailed analysis of its performance. The methodology for this paper we are using is the concept of bloom filter as Bloom filters are used to help identify transactions that are unknown to a node or in a block. Our proposed system will check if a specific transaction is already in the Ethereum transaction mempool? If yes, then the transaction is dropped and not forwarded to peers. If it is the first time that the node is seeing the transaction, it is added to the Ethereum bloom filter mempool and forwarded to peers. Similarly, transactions can be removed from the Ethereum mempool by removing the transaction from the Ethereum bloom filter mempool. Since, the bloom filter and its variants are probabilistic structures, telling us whether TX is in the mempool or definitely not in the mempool. We also evaluate the impact of design parameters such as bloom filter and counting bloom filter dimensions, on performance metrics (space-time trade off and accuracy). We implement our model in C++ and perform a simulation on a month of Ethereum logs which comprises about a million transactions and analyze the performance of OMCEM.