Hazelcast, the open source in-memory data management company with tens of thousands of installed clusters and over 46 million server starts per month, announced the 0.6 release of Hazelcast Jet – the ultra fast, application embeddable, stream and batch processing engine.
New functionality in Hazelcast Jet 0.6 includes a simplified high-level Java API, enhanced job management capabilities which enable computation elasticity and Spring integration. Hazelcast Jet is a 3rd generation big data processing engine, the first two generations being Hadoop and Spark. With performance up to 15x faster than Spark combined with great simplicity, Hazelcast Jet is setting a new standard.
In its 0.5 release, Hazelcast Jet launched its Pipeline API. In Hazelcast Jet 0.6, the Pipeline API has been expanded to fully cover all Hazelcast Jet features. In 0.6, it is even more straightforward to define sliding, tumbling and session windows or to deal with late events.
The API was streamlined to provide convenience for integrating 3rd party libraries such as predictive analytics tools or enrichment, users will find it easy to write a custom sink to output data to any system of record. Stream processing has overtaken batch processing as the preferred method of processing big data sets and these new features are designed to satisfy the needs of companies that require immediate insight into their data.
To enable users to fully understand the most current capabilities of Hazelcast Jet, 8 new demo applications have been developed which illustrate how to use Hazelcast Jet to solve real-world problems. Examples include applications for Real-Time Image Recognition, Twitter Cryptocurrency Sentiment Analysis, Real-Time Road Traffic Analysis and Prediction, Real-Time Sports Betting Engine, Flight Telemetry, Markov Chain Generation and Market Data Ingest.
The new elasticity features of Hazelcast Jet 0.6 allow for elastic scale out and scale in, perfect for the cloud. Executing jobs can be rescaled to make use of new nodes as they added – meaning running production jobs will scale to utilise computational resources as and when it’s made available, even while in mid-production. When a node is removed, computations are resized to the smaller cluster without interruption.
Hazelcast Jet 0.6 is easier than ever to use. Anyone who is a Java 8 programmer will find it very natural and simple. To accommodate the large audience of Spring users, Spring integrations are added. In Hazelcast Jet 0.6, a Jet client or Jet instance can be configured as a Spring Bean and developers are now able to autowire them with their applications. Also, the Spring Context of the Jet cluster can be injected directly into a Jet job.
Additional enhancements in Hazelcast Jet 0.6 include:
- Updated Hazelcast Jet connectors – Kafka v1.0 and HDFS v2.8.3
- Embedded Hazelcast IMDG updated to v3.10
- Entry Processor Sink – when writing to IMap, Hazelcast IMDG users can now update just a part of the entry instead of rewriting the complete entry.
- Java 9 – Hazelcast Jet builds and runs on Java 9 and supports the new Java module system.
Greg Luck, CEO of Hazelcast, said: “Jet is being adopted rapidly by the Global 200. The deployment and programming simplicity, its embeddability, and the validated performance advantage of Jet are driving this. Hazelcast Jet 0.6 gains valuable features for even broader deployment.”
Hazelcast Jet is simple to program and deploy – and can be fully embedded for OEMs and microservices – making it simpler and easier than ever before to build next generation systems.