LMCache provides open-source KV cache infrastructure for LLM inference, including persistent tiered cache storage, reuse across serving engines, operational HTTP APIs, a frontend dashboard, observability, and production deployment patterns for Docker and Kubernetes.