A Dataflow streaming job writes aggregates to a Basic Tier (4GB) Memorystore for Redis used by ~40 read‑only clients. Expecting hundreds of read‑only clients, ensure read/write availability and fast deployment. What should you do?

The correct option addresses the need for high availability (HA), increased read capacity, and fast deployment. A Standard Tier Memorystore for Redis instance provides HA with automatic failover. Adding multiple read replicas significantly increases read throughput, accommodating hundreds of read-only clients without impacting write performance. Increasing the instance size to 5 GB ensures sufficient capacity for potential data growth. The other options are less suitable: Creating a Standard Tier instance with no read replicas (first option) provides HA but doesn't scale read capacity for hundreds of clients. Replacing with Memorystore for Memcached (third option) would require significant application changes (Dataflow and clients) and doesn't offer the same data structures or persistence as Redis, making it a more complex and less direct solution. Creating multiple Basic Tier instances (fourth option) complicates client logic for sharding and doesn't inherently provide HA or automatic failover for individual instances.