Which of the following characteristics is most indicative of a NoSQL database's focus on scalability and availability over strict consistency?

NoSQL databases are designed for high scalability and availability, which is primarily achieved through a distributed architecture. Partitioning, also known as sharding, distributes data across multiple server nodes, enabling horizontal scaling and preventing any single machine from becoming a bottleneck. Replication creates copies of data on different nodes. This redundancy ensures that the system remains available and operational even if some nodes fail. This architectural choice directly supports scalability and availability, often at the expense of the strict, immediate consistency found in traditional relational databases, favoring an "eventual consistency" model instead.

A. Complex joins and multi-table transactions are hallmarks of relational databases, often sacrificed in NoSQL systems for performance and scalability.

B. Strict referential integrity is a consistency mechanism typical of relational databases, which NoSQL often relaxes for higher availability.

D. A centralized, single-instance architecture is the opposite of the distributed model used by NoSQL for scalability and availability; it represents a single point of failure.

1. Oracle NoSQL Database Administrator's Guide, 23.3: In the "Designing a Store" section, it states, "The primary reason for sharding a store is for scalability... Replication is used to protect against node failure and to provide read scalability." This directly links partitioning (sharding) and replication to scalability and availability.

2. Oracle NoSQL Database Concepts, 23.3: The "High Availability" chapter explains, "Oracle NoSQL Database is a shared-nothing, replicated, scale-out system... Data is replicated for redundancy and high availability." This confirms replication as a core mechanism for availability.

3. Brewer, E. (2012). "CAP Twelve Years Later: How the 'Rules' Have Changed." Computer, 45(2), 23-29. https://doi.org/10.1109/MC.2012.37. This foundational academic paper discusses the CAP theorem (Consistency, Availability, Partition tolerance), explaining that distributed systems like NoSQL databases must often choose between strong consistency and high availability when partitions occur. The use of partitioning and replication is central to this trade-off.