Hazelcast 3.6 Introduces New Cloud and Container Deployment Options and Unprecedented Vertical Scalability to 10s of TBs of RAM

Version 3.6 Establishes New Standards for Super-fast Persistence and Performance

News Highlights:

  • Hazelcast® Version 3.6 includes numerous performance improvements to give industry leading performance verified with published benchmark suites versus Oracle Coherence, Red Hat JBoss Data Grid/Infinispan and GridGain/Apache Ignite.
  • New Hot Restart Store adds persistence with write performance of 310,000 writes per second per node and restart times of 7 seconds per 1.3GB per node (with parallel load across nodes).
  • High-Density Memory Store (HDMS) is extended to include JCache, Map, Hibernate and Web Sessions data structures to allow the creation of huge caches with hundreds of GBs per node.
  • Cloud enhancements include new container deployment options for Docker, and support for Kubernetes, Apache jclouds, Consul, CloudFoundry and OpenShift PaaS.

Palo Alto, Calif., 21 January, 2016Hazelcast, the leading provider of in-memory data grids and large-scale enterprise caching with tens of thousands of installed clusters and over 13 million server starts per month, today announced the general availability of Hazelcast 3.6. The latest version includes performance enhancements which make Hazelcast the fastest in class IMDG, as demonstrated by published benchmarks: Oracle Coherence Benchmark Suite, Red Hat Infinispan Benchmark, Pivotal Gemfire/Apache Geode Benchmark and GridGain/Apache Ignite Benchmark. Hazelcast 3.6 also contains new cloud management and container deployment options that make it easier for users to set-up, configure and manage cloud or hybrid on-premise deployments, and Hot Restart Store – a new high-performance persistence feature.

With the Hot Restart Store mission-critical e-commerce and enterprise applications can start up hot – with their data, without having to reload data from a system of record. Start-up times are very fast: 1.3GB/per second per node for SSD. While operating it supports 310,000 writes per second per node, so that the operation of the system is not slowed down. This industry-leading reboot performance provides a safety backup in case all the nodes in the cluster crash (data center failure). Additionally, software updates are now possible without having to reload data. Hot Restart Store will support JCache, Map, Hibernate and Web Sessions data structures.

Hazelcast 3.6 also expands High-Density Memory Store (HDMS) capabilities for JCache, Map, Hibernate and Web Sessions data structures enabling memory caches for the largest scale-out or scale-up of application deployments. HDMS enables each node to hold very large amounts of data, 100s of GBs per node, without incurring garbage collection overhead, as well as allowing applications to Near Cache hundreds of GBs of data at a time.

The Hazelcast Cloud Discovery Service Provider Interface (SPI) enables cloud-based or on-premise nodes to auto-discover each other with minimum configuration and management set-up. 3.6 comes with a multitude of providers: Kubernetes, Apache jclouds, Consul, etcd, and a new AWS provider.

Hazelcast 3.6 also includes new container deployment options for Docker, along with support for PaaS providers OpenShift and CloudFoundry.

In version 3.6, Hazelcast also announces the availability of its Open Binary Client Protocol in a move to further support community development of new clients. Already community members have developed clients for Python, Ruby, NodeJS, Scala and Clojure which are all freely available as open source. Additionally, Hazelcast’s native clients for C++ and .NET are now open source giving developers more tools for new client API development. Critically, Hazelcast is now a step closer to becoming platform-independent as developers are now able to write clients in the program language of their choice.

Greg Luck, CEO of Hazelcast, said: “We are really excited about 3.6. It is the biggest release in two years. HD is now available across the board. We are fully geared up for new cloud deployments. We now have clients for 8 different programming languages. And we are better than ever at dealing with huge caches. With this release we have become a platform and our amazing community has already started work extending Hazelcast like never before.”