Normalisation in relational databases is a process designed to reduce data redundancy and improve data integrity. It involves organizing data in a database into tables and defining relationships between them according to rules designed to safeguard the data and make the database more efficient. Normalisation typically involves several stages, each termed as a 'normal form'. These stages include eliminating duplicate columns from the same table (First Normal Form), creating separate tables for sets of values that apply to multiple records (Second Normal Form), and removing columns not dependent on the primary key (Third Normal Form). By following these normalisation rules, relational databases ensure that data is stored logically and efficiently. This reduces the likelihood of data inconsistencies and anomalies, thus enhancing data integrity. Normalisation also simplifies database maintenance and improves database performance by organizing data into well-defined structures.

Foreign keys in relational databases play a critical role in maintaining referential integrity and establishing relationships between tables. A foreign key is a field in one table that uniquely identifies a row of another table. Essentially, it is a reference to the primary key in another table. This linkage ensures that the data in the two tables remains consistent and accurate. For instance, if Table A has a foreign key that references the primary key of Table B, it ensures that each entry in Table A corresponds to an existing entry in Table B. This prevents orphan records and maintains the logical connection between related data entities. Foreign keys are crucial in enforcing data integrity and are a fundamental aspect of relational database design, ensuring that relationships between tables are correctly represented and maintained.

While relational databases are widely preferred for their efficiency and data integrity, there are scenarios where a file-based system might be more appropriate. One such scenario is when dealing with simple, small-scale applications where the overhead of setting up and maintaining a relational database is not justified. File-based systems can be more straightforward and less resource-intensive for basic data storage needs. Another scenario is when the application requires very fast read and write access to a small amount of data, and the complexity of SQL queries and relational database management would be an unnecessary complication. Additionally, for applications that do not require complex data relationships, constraints, or high levels of data integrity, a file-based system might suffice. In these cases, the simplicity and low overhead of file-based systems offer practical benefits over the more complex and resource-intensive relational databases.

Relational databases are primarily designed for structured data, organized into defined tables and fields. However, they can handle unstructured data to some extent through the use of specific data types and techniques. For instance, large text fields (like VARCHAR) or Binary Large Objects (BLOBs) can be used to store unstructured data such as text documents or multimedia files. Modern relational database systems have also introduced features like JSON and XML data types to store and query unstructured or semi-structured data. While these methods enable the storage of unstructured data, it's important to note that relational databases may not be as efficient in processing unstructured data compared to specialized databases like NoSQL, which are explicitly designed for such purposes. Nevertheless, the ability to handle a mix of structured and unstructured data makes relational databases versatile and adaptable to various data management needs.

Relational databases effectively manage large-scale data while maintaining performance through a combination of indexing, query optimization, and efficient storage mechanisms. Indexing, similar to an index in a book, allows the database to quickly locate and retrieve data without scanning the entire table. This significantly speeds up query processing, especially in large datasets. Query optimization involves the database engine analysing different query execution plans and selecting the most efficient one. This process is crucial in handling complex queries efficiently. Additionally, relational databases often use advanced storage techniques like partitioning, where data is divided into smaller, manageable segments, reducing the load on any single storage component and improving performance. These methods, combined with the robust architecture of relational databases, enable them to handle vast amounts of data while ensuring swift data access and manipulation.