In the burgeoning realm of blockchain technology, the proliferation of decentralized finance (DeFi) has been accompanied by various malicious activities, such as sandwich attacks, 51% attacks, and Sybil attacks. Sandwich attacks threaten the integrity and fairness of cryptocurrency transactions by leveraging the transparency of public mempools and the transaction order dependency in automated market maker (AMM) mechanisms. Attackers strategically manipulate asset prices to extract value from targeted transactions. Although many countermeasures have been proposed, existing defenses are still fragmented across different technical domains. Moreover, a comprehensive evaluation of their effectiveness and limitations is lacking, making it difficult to assess their overall impact. To address this gap, we introduce a novel taxonomy that systematically categorizes defense strategies throughout the transaction lifecycle, spanning from user parameter configuration to blockchain confirmation. Meanwhile, we conducted a comprehensive review of existing defense strategies against sandwich attacks, critically analyzing their strengths and limitations. Our findings indicate that while these strategies have shown promising results in mitigating conventional sandwich attacks across various dimensions, significant limitations remain. Hence, this article aims to help researchers address the sandwich attack issues from the perspective of the transaction lifecycle.