Replication And High Availability Clustering

via “distributed-cluster-deployment-with-peer-replication”

The AI-native database built for LLM applications, providing incredibly fast hybrid search of dense vector, sparse vector, tensor (multi-vector), and full-text.

Unique: Implements peer-to-peer replication with Thrift RPC for vector databases, enabling horizontal scaling without central coordinator; uses C++20 modules for compile-time cluster protocol optimization and lock-free synchronization primitives.

vs others: More decentralized than Milvus (which uses Etcd for coordination) because Infinity uses peer-to-peer Thrift; simpler than Elasticsearch clustering because Infinity's replication model is optimized for append-heavy vector workloads.