In ERDs, cardinality represents the number of instances of an entity that can be associated with another entity. It defines how entities are related and the possible quantity of relationships, such as one-to-one, one-to-many, or many-to-many.

The primary objective of normalization in database design is to minimize data redundancy and dependency. This ensures that data is efficiently organized, reduces the risk of anomalies, and enhances data integrity in the database.

Storage optimization in relational schema design focuses on enhancing query performance by organizing and storing data efficiently. It involves strategies like indexing, partitioning, and denormalization to ensure quick and effective data retrieval.

Compression typically has no impact or can even improve data retrieval speed in a database system. By reducing the amount of data that needs to be transferred, it can enhance overall performance. However, the specific impact depends on the type of compression algorithm used and the characteristics of the data.

A potential drawback of partitioning a database is the complexity in administration and maintenance. While partitioning can enhance performance and simplify certain operations, managing and maintaining partitions can become complex, especially as the database scales. This requires careful planning and monitoring.

The primary purpose of indexing in a database is to speed up data retrieval. Indexing allows the database system to locate and access the required data more quickly, improving overall query performance.

The correct term is Functionally Dependent. An attribute is functionally dependent on the primary key if its value is uniquely determined by the entire primary key, not just a proper subset. Understanding this concept is crucial for database design and normalization.

To optimize performance in a Key-Value Store, all of the above techniques may be employed. Caching helps reduce the need to fetch data from the underlying storage repeatedly, indexing improves lookup speed, and compression reduces the amount of data transferred, collectively enhancing overall system performance.

Type 3 SCD is advantageous as it captures both current and historical data. It maintains a separate column for historical changes, providing a snapshot of changes over time without excessive storage requirements. This is particularly useful for scenarios where a limited history is needed.

Composite indexing is an approach where the index is built on multiple columns. This technique is beneficial for optimizing query performance, especially when queries involve those specific columns.