Features
Infinispan is designed with ease of use and performance in mind but also gives you enterprise-grade capabilities for a wide range of use cases.
Flexible Deployment Models
- Embed as a library for in-process caching with zero network overhead.
- Deploy as a standalone server for language-independent remote access.
- Run as a container image with Docker, Podman, or any OCI runtime.
Designed for the Cloud
- Lightweight, cloud-native design for microservice architectures.
- Kubernetes Operator that provides operational intelligence.
- Container image that you can run anywhere: OpenShift, AWS EC2, Azure, GCP.
Clusterable and Elastic
- Automatically form clusters with configurable discovery mechanisms.
- Elastically scale by adding or removing nodes with automatic data rebalancing.
- Create global clusters that replicate data across multiple data centers.
Vector Search & AI
- Distributed vector indexes with kNN similarity queries for embeddings.
- Float and byte vector support for flexible embedding storage.
- Semantic search, RAG, and semantic caching for LLM applications.
- Native integration with Spring AI, LangChain4j, LangChain, and Quarkus LangChain4j.
Search
- Ickle query language allows relational and full-text search with straightforward syntax.
- Protobuf-based schema indexing for structured, type-safe queries.
- Continuous queries that push results to your application as data changes.
Configurable data storage
- Configure cache modes to distribute and replicate data as needed.
- Encode key/value entries with different storage formats.
- Store data in the JVM heap or native "off-heap" memory.
Expiration and Eviction
- Automatically manage memory with TTL (time-to-live) and max-idle expiration.
- Evict entries based on entry count or memory usage to prevent out-of-memory errors.
- Touch-based idle expiration keeps frequently accessed data warm.
Persistent Cache Stores
- Store data both in memory and durable persistent locations.
- Cache stores available for File, RocksDB, JDBC, and more out of the box.
- Use write-through caching to eliminate expensive data requests for applications.
HotRod Remote Protocol
- Language-independent, binary protocol for high-performance remote access to data.
- Client intelligence dynamically responds to cluster topology changes and requests data from primary owners.
- Near-caching on the client side for ultra-low latency reads.
- Implementations available in Java, C++, C#, Python, and Node.js.
Developer Tooling
- Quarkus extensions and Spring Boot starter for rapid development.
- Testcontainers module for integration testing with real Infinispan instances.
- APIs for programmatic usage and extensibility.
- Integration with Hibernate ORM, JCache, Micronaut, and WildFly.
Secure
- Control access to data based on user roles and permissions.
- Define server security realms that protect you from intrusion and attack.
- Encrypt inter-cluster traffic and remote endpoints.
Administration and Monitoring
- Infinispan Console gives you intuitive administration and monitoring capabilities.
- Infinispan CLI provides simple, yet powerful, tooling for administration tasks
- Get statistics and MicroProfile compatible metrics for observability over time.
ACID Transactions
- Ensure data integrity with distributed transactions.
- Configure transactions with optimistic or pessimistic locking, synchronization, and different isolation levels.
- XA transactions with full recovery available.
Event Listeners
- Deploy local or clustered listeners to detect events.
- Get notifications when changes occur to the data store or Infinispan clusters.
Code execution
- Execute distributed operations across the cluster for parallel data processing.
- Aggregate and transform data with cluster-wide compute tasks.
