The traditional approach to database engineering has long been based on the ACID (Atomicity, Consistency, Isolation, Durability) model. While this model has served the industry well for several decades, modern databases have evolved to require more flexibility and scalability, thus leading to the development of the BASE (Basically Available, Soft-state, Eventually consistent) model.

The ACID model provides strong guarantees to ensure data integrity, but it also comes with several limitations. One of the major drawbacks of ACID is its focus on consistency. This means that if a transaction fails at any point, the entire operation will be rolled back, and the database will be reverted to its previous state. This approach can be problematic in modern applications that require high availability and real-time processing.

On the other hand, the BASE model makes some trade-offs to provide greater scalability and availability. Instead of prioritizing strong consistency, BASE allows for eventual consistency. This means that the database may not always be in a fully consistent state, but it will eventually reach that state after all transactions have been processed. This approach allows for greater scalability and availability, as transactions don’t need to wait for each other to complete before proceeding.

In addition to eventual consistency, the BASE model also emphasizes availability and soft-state. Availability refers to the ability of the database to remain accessible even if some nodes in the distributed system fail. Soft-state, on the other hand, refers to the ability of the system to tolerate failures and recover gracefully.

While the ACID model is still widely used, it is becoming increasingly clear that modern databases require more flexibility and scalability than ACID can provide. In fact, many of the most popular databases today, such as Apache Cassandra and MongoDB, are built on the principles of the BASE model.

The move towards BASE does not mean that consistency is no longer important. Instead, it means that consistency is achieved over time, rather than being immediately enforced. This approach can be beneficial in many scenarios, such as when processing large volumes of data in real-time or when dealing with distributed systems that require high availability.

In conclusion, abandoning ACID in favor of BASE in database engineering is a major shift in the industry, but one that is necessary as databases continue to evolve. While the ACID model provided strong guarantees for data consistency, the BASE model provides greater flexibility and scalability, allowing modern databases to perform at their best. As database architects, it is important to understand the trade-offs of each model and to select the right approach for each use case.