A media firm stores 500 TB in us-east-1 S3 and DynamoDB session tables, must replicate only EU customer prefixes and active sessions to eu-west-1 while minimizing transfer and storage spend. Which cost-aware multi-Region replication design meets resilience with lowest cost?

The correct design is right because S3 Cross-Region Replication with prefix filters copies only the in-scope EU prefixes to eu-west-1 on write, avoiding transfer and storage charges for the remaining non-EU data within the 500 TB bucket, while DynamoDB global tables provide active-active low-latency replication, and a TTL that expires ended sessions keeps only active sessions stored and replicated (global tables cannot filter items themselves), which together deliver multi-Region read resilience with the lowest incremental cost. A design pairing prefix-filtered replication with DynamoDB point-in-time recovery fails on recovery dynamics because backup and on-demand restore is not continuous replication and cannot serve live session reads in eu-west-1 without RTO and manual restore work. A design using S3 Batch Replication for EU prefixes fails on continuity because batch handles existing objects as a one-time backfill and does not replicate new writes, so ongoing EU ingest would diverge without continuous rules. A design that adds global tables for full session history fails on cost efficiency because replicating expired and historical items multiplies write, transfer, and storage spend contrary to the explicit active-sessions-only constraint. Exam caveat: distinguish continuous replication for resilience from backup or migration copy services that sound similar but have different RPO and cost models. Practical check: verify S3 replication metrics for filtered bytes, confirm the TTL attribute expires ended sessions so only active keys remain, and estimate monthly cross-Region transfer plus replica storage before enabling.