A large e-commerce platform experiences sudden spikes in traffic during sales events. How would you architect the database infrastructure using NoSQL databases to handle this scalability requirement?

A distributed database like Cassandra or MongoDB with sharding and replication is suitable for handling sudden traffic spikes as it can scale horizontally by adding more nodes. Sharding distributes data across multiple nodes, while replication ensures data availability and fault tolerance. A caching layer improves read performance but may not handle spikes in write operations efficiently. Message queues are used for asynchronous communication and may not directly address scalability. Load balancers distribute incoming traffic across multiple servers but do not inherently handle database scalability.