Step 1: Recall why NoSQL databases were created.
As applications began dealing with huge, fast growing, and often unstructured data, traditional relational databases with their rigid schemas and single server scaling started to struggle, NoSQL systems were built specifically to address these limitations.
Step 2: Check each option against that motivation.
A fixed schema is exactly what relational databases enforce, not NoSQL, which typically allows flexible or schema less structures. The relational data model with tables and joins also describes SQL databases, not NoSQL, which instead uses key value, document, column family, or graph models. Claiming NoSQL only supports structured data is backwards too, since it is prized for handling unstructured and semi structured data as well.
Step 3: Identify the trait that actually defines NoSQL.
What genuinely defines NoSQL databases is their ability to scale out horizontally across many servers and their flexibility in how data is modeled and stored, which is exactly what makes them suitable for large, fast changing datasets.
Step 4: Conclude.
The key characteristic of NoSQL databases is
\[ \boxed{\text{Scalability and flexibility}} \]