What is a direct consequence of relying solely on embedded documents in MongoDB?

When relying solely on embedded documents in MongoDB, one direct consequence is the potential for worsening read and write performance under large datasets. This happens because embedded documents increase the size of the parent document. In cases where the embedded documents grow significantly, it can lead to situations where the parent document exceeds the BSON document size limit (16 MB).

Additionally, since MongoDB stores documents as single entities, large embedded documents can lead to increased memory usage and affect the efficiency of read and write operations. If queries often need to extract information from deeply nested or large embedded documents, this can slow down query performance as more data needs to be processed and transferred.

While there are benefits to using embedded documents, such as improved data locality and sometimes better performance for certain query patterns, using them excessively or without considering the size implications can lead to negative impacts on performance as the dataset grows.