Ultra-Fast Processing Framework

Stream and Batch Processing

Use data instantly.

The Hazelcast Platform provides event stream processing and fast batch processing for any scale.

It extracts live data from databases, data lakes, applications, devices, and message brokers such as Apache Kafka, Apache Pulsar, AWS Kinesis, or RabbitMQ and converts raw, high-volume data streams to business events and actionable insights convenient for consumption by applications, dashboards, and databases.

Serves applications

Hazelcast is the only streaming platform that comes with integrated distributed storage to serve processed live data to applications. Access it through thousands of concurrent low-latency queries or push updates to the apps in an event-driven fashion.

Keeps up with the real-time data without losing the momentum!

Hazelcast minimizes the impact of massive amounts of data to lower the end-to-end processing latency. It stays consistently fast, relieving the load from downstream systems. A single node of Hazelcast has been proven to crush 25 million events per second with latency constantly under 10 milliseconds. It also scales nearly linearly to process 1 billion events per second on only 720 cloud virtual CPUs.

Use Cases

React to real-time events for better situation awareness and faster, more personalized decisions.

Detect complex patterns, use statistical functions and apply algorithmic analytics such as ML model inference.

Processing infinite, out-of-order data in the distributed setup can be hard. Reduce the programming and operational complexity by using high-level APIs and let Hazelcast do the heavy lifting.

Read the Case Study

Integrate high-frequency streaming data with your applications. Hazelcast relieves load from downstream systems by converting raw, high-volume data to business events convenient for consumption by applications and databases.

Combine live event streams with historical or operational data.

Store pre-processed, aggregated data to Hazelcast for fast consumption. Or push it to an analytics database, index it with a search engine or visualize it on a dashboard.

Watch Our Flight Telemetry Demo

Build and maintain derived data views from the business events.

Hazelcast pulls events from a message bus or transparently extracts them from your applications using non-intrusive change data capture.

Correlate events from multiple sources, maintain domain-specific data views.

Push processed events to subscribed applications or serve it using OOTB fast data layer.

Hazelcast can connect to a variety of data platforms with its portfolio of out-of-the-box connectors, and read the data in parallel at high speeds to process the data in a large-scale, batch manner. The Hazelcast performance advantage lets you transform, aggregate, filter, enrich, and score data faster than your traditional batch processing tools.

You can also store the processed data into the Hazelcast in-memory data store to boost the speeds of additional, downstream processing jobs.

Hazelcast Features

Resilient to Failures

Hazelcast provides correct results even if it experiences failures such as hardware or network errors or outages of connected systems. It utilizes distributed transactions and distributed snapshots to provide exactly-once end-to-end processing guarantees under the face of failures. It supports multi-datacenter deployment for disaster recovery scenarios.

Proven to Handle Massive Load without Slowing Down

Hazelcast can achieve extremely low latencies while processing millions of items per second on just a single node. Cluster performance can be tuned without downtime by adding and removing resources on the fly. More Info

Advanced Event Processing

Event data can often arrive out of order due to multiple data sources or network routing. Hazelcast implements advanced techniques to suppress negative impact of out-of-order data and to produce correct results.

Works with In-Memory Computing

Hazelcast in-memory computing is highly-concurrent data storage with low-latency queries and fine-grained, key-based access and eventing. Hazelcast stream processing cooperates with in-memory computing to distribute processed data. Pre-existing Hazelcast clusters can be upgraded to add stream processing.

Enterprise Operations

Enterprise comes with comprehensive tooling for monitoring, cluster and job management, deployment, version control and integrated security for convenient, zero-downtime operations. To eliminate the operation complexity, it does not require the presence of any other software infrastructure (such as Hadoop, Kafka or ZooKeeper) to run, except Java.

Container and Cloud Ready

Hazelcast is a CNCF-listed project. It uses Operators to run and manage applications as cloud-native in Kubernetes and OpenShift cluster environments. Variety of plugins enables deploying a cluster to a public clouds (AWS, Azure, GCP).
More Info


Hazelcast integrates with many systems and applications producing and consuming data such as message brokers, search engines, filesystems, RPC services or databases including the change data capture. See the full list.

Rooted in Open-Source

Hazelcast is an open-source project with commercial extensions and enterprise-level support. Compare the features.

Who Uses Hazelcast Stream Processing







BNP Paribas

BNP Paribas







Flow Traders

Flow Traders

Plex: The Manufacturing Cloud

Plex: The Manufacturing Cloud



Get Started with Hazelcast

See how easy it can be to add Hazelcast to your business. Start your free 30-day trial now.

Featured Resources


Stream Processing Essentials

The basics of stream processing using Hazelcast Jet