LIMAO: A Framework for Lifelong Modular Learned Query Optimization

In traditional database systems, query optimizers play an important role in ensuring efficient query execution. Recently, learned query optimizers (LQOs) have shown promising improvements over traditional ones. However, most existing LQOs assume a static query environment (or limited dynamic environments), which limits their ability to handle real-world scenarios where queries and workloads change dynamically. This limitation leads to performance degradation over time, as frequent retraining can cause the model to forget previously learned knowledge, a problem known as catastrophic forgetting. In this research, we propose a novel approach to address this issue.