Infinispan is an open source data grid platform. It exposes a JSR-107 compatible Cache interface (which in turn extends java.util.Map) in which you can store objects. While Infinispan can be run in local mode, its real value is in distributed mode where caches cluster together and expose a large memory heap. Distributed mode is more powerful than simple replication since each data entry is spread out only to a fixed number of replicas thus providing resilience to server failures as well as scalability since the work done to store each entry is constant in relation to a cluster size.
So, why would you use it? Infinispan offers:
Massive heap and high availability - If you have 100 blade servers, and each node has 2GB of space to dedicate to a replicated cache, you end up with 2 GB of total data. Every server is just a copy. On the other hand, with a distributed grid - assuming you want 1 copy per data item - you get a 100 GB memory backed virtual heap that is efficiently accessible from anywhere in the grid. If a server fails, the grid simply creates new copies of the lost data, and puts them on other servers. Applications looking for ultimate performance are no longer forced to delegate the majority of their data lookups to a large single database server - the bottleneck that exists in over 80% of enterprise applications!
Scalability - Since data is evenly distributed there is essentially no major limit to the size of the grid, except group communication on the network - which is minimised to just discovery of new nodes. All data access patterns use peer-to-peer communication where nodes directly speak to each other, which scales very well. Infinispan does not require entire infrastructure shutdown to allow scaling up or down. Simply add/remove machines to your cluster without incurring any down-time.
Data distribution - Infinispan uses consistent hash algorithm to determine where keys should be located in the cluster. Consistent hashing allows for cheap, fast and above all, deterministic location of keys with no need for further metadata or network traffic. The goal of data distribution is to maintain enough copies of state in the cluster so it can be durable and fault tolerant, but not too many copies to prevent Infinispan from being scalable.
Persistence - Infinispan exposes a CacheStore interface, and several high-performance implementations - including JDBC cache stores, filesystem-based cache stores, Amazon S3 cache stores, etc. CacheStores can be used for "warm starts", or simply to ensure data in the grid survives complete grid restarts, or even to overflow to disk if you really do run out of memory.
Language bindings (PHP, Python, Ruby, C, etc.) - Infinispan offers support for both the popular memcached protocol - with existing clients for almost every popular programming language - as well as an optimised Infinispan-specific protocol called HotRod. This means that Infinispan is not just useful to Java. Any major website or application that wants to take advantage of a fast data grid will be able to do so.
Management - When you start thinking about running a grid on several hundred servers, management is no longer an extra, it becomes a necessity. The preferred way to manage multiple Infinispan instances spread accross different servers is to use RHQ which is JBoss' enterprise management solution. Thanks to RHQ's agent and auto discovery capabilities, monitoring both Cache Manager and Cache instances is a very simple.
Support for Compute Grids - Infinispan 5 adds the ability to pass a Runnable around the grid. This allows you to push complex processing towards the server where data is local, and pull back results using a Future. This map/reduce style paradigm is common in applications where a large amount of data is needed to compute relatively small results.