To answer your questions:
1. The library itself is stateless, so it scales along with backend services.
2. The size is limited by backend service, such as the size of your cache cluster.
3. Because all operations go through the backend service, the existing telemetry, logging and monitor tools are still applicable. However it cold be a good idea to include more metrics counters to provide fine-tuned perf info.
4. There will be more blog articles and video introductions coming out. Keep tuned!