In-Memory Distributed Database
What is Infinispan?
Infinispan is an open-source in-memory database that can hold nearly any type of data, from plain-text to structured objects. Retrieve your data with lightning-fast speeds with Infinispan's powerful full-text and vector search capabilities. Guarantee reliability and integrity by distributing your data across elastically scalable Infinispan clusters that offer high availability, fault tolerance, and the ability to replicate between multiple geographic sites. Connect to Infinispan using either a native, high-performance protocol or a Redis or Memcached client of your choice.
Ready to start using Infinispan?
Get Started NowFeatures
Interoperability
Access data across multiple protocols and programming languages.
Resilient and Fault Tolerant Data
Ensure data is always available to meet demanding workloads.
ACID Transactions
Guarantee that data is always valid and consistent.
Clustered Processing
Process data in real-time without burdening resources.
Queries
Perform simple, accurate, and fast searches across distributed data sets.
Boost Application Performance
Infinispan turbocharges applications by storing data closer to processing logic, which reduces latency and increases throughput.
Available as a Java library, you simply add Infinispan to your application dependencies and then you’re ready to store data in the same memory space as the executing code.
If you want to provision a data layer that is independent of your applications, you can use Infinispan Server for remote access to data with in-memory performance. Clients are a single network hop away from data through consistent hashing techniques and can make requests over HTTP or with a custom binary TCP protocol called Hot Rod.
Learn MoreAchieve High Availability and Elasticity
Infinispan provides trusted open-source technology to deliver scalability to meet workload demands and reduce resource utilization. At the same time, Infinispan distributes your data across clusters so no single point of failure causes data loss.
One popular use for Infinispan is as a shared store for stateful data, such as user HTTP sessions. Applications can stay lightweight and avoid heap usage by externalizing sessions to Infinispan clusters, which act as an independent data layer.
Learn More
Backup Across Data Centers
Infinispan clusters running in different geographical locations can form global clusters to back up your data across sites. If sites go offline clients can immediately switch to an available cluster, making sure data center faults do not cause service interruptions.
When using the Infinispan Operator with Kubernetes environments such as Red Hat OpenShift, cross-site replication capabilities make your data ready for hybrid and multi cloud deployments.
Infinispan also guarantees data consistency when using cross-site replication, even in cases where clients make concurrent writes at different locations that use asynchronous replication. So your data is always there and always accurate, no matter where you’re running.
Learn MoreFeatured Use Cases
Quarkus extensions
Spring and Spring-Boot
Redis Hot Replacement
Keycloak
Langchain
Langchain4j
Hibernate
Apache Camel
Micronaut
Wildfly
Vert-x
Kogito
Recent Blog Posts
December 18, 2025
By Tristan Tarrant
We don’t usually blog about micros… but this is the second time in as many days that I have to break with tradition. Serious ProtoStream bug with Latin1 strings If you were using strings that could be represented as Latin1/ISO-8859-1 in your protoco...
December 17, 2025
By Tristan Tarrant
We don’t usually blog about micros, unless there are important things to mention, and in this case we have a couple of significant changes to share with you. Spring Framework 7 and Spring Boot 4 support Spring Framework 7 and Spring Boot 4 were released just aft...
November 10, 2025
By Tristan Tarrant
"Keep Rollin'" Infinispan 16.0 is here, and it is codenamed "Keep Rollin'". This is a very significant release, as it introduces zero-downtime in-place rolling upgrades: the capability of incrementally upgrading individual nodes in a cluster without having to stop opera...
