Design a scalable system for a token-generation service used by an LLM that needs to handle up to 100,000 requests per second.

Focus on outlining the architecture of a scalable token-generation service that can efficiently handle high throughput. Start by discussing the choice of technology stack, including programming languages and frameworks that support concurrency and high performance. Detail how you would implement load balancing to distribute requests evenly across multiple servers, and consider using microservices for modularity. Address data storage solutions, such as in-memory databases for fast access, and discuss caching strategies to reduce latency. Finally, touch on monitoring and scaling strategies, including auto-scaling mechanisms to adapt to varying loads.