In the world of databases, consistency is often misunderstood and can be a major challenge for NoSQL databases. The CAP theorem highlights three types of consistency: strong consistency, timeline consistency, and eventual consistency. Strong consistency ensures that all writes are acknowledged and read requests always see a complete, consistent result. Timeline consistency allows some recent data to be lost in case of failures, while eventually consistent databases return inconsistent data temporarily. Common myths about eventual consistency include the idea that implementing quorum reads and writes can make it strongly consistent, which is incorrect; that it delivers higher performance than strongly consistent databases, which is also incorrect; and that it's easy to scale out, which is not true. Ultimately, a database with strong consistency, efficiency in scaling out, high performance, low latency, high availability, and high data density can be achieved through proper architecture.