Which solution meets these requirements MOST cost-effectively?

Explanations:

Amazon EMR is typically used for batch processing and is not a direct solution for scaling Redshift to handle spikes in complex queries. It would add extra cost and complexity without directly addressing Redshift capacity.

A classic resize operation is time-consuming and would cause downtime, which does not meet the requirement of handling bursts while maintaining availability for read and write operations.

Elastic resize in Redshift can be quicker than classic resize but is still not instantaneous and may impact availability temporarily, failing to meet the requirement of uninterrupted read and write access.

Concurrency Scaling is designed to handle unexpected bursts by automatically adding transient capacity when needed, ensuring high availability for read and write operations with cost-effective pricing for sporadic spikes.