Build Data Pipelines That Scale
Build Data Pipelines That Scale
Jet consumes and analyzes millions of events per second or terabytes of data at rest using a unified API. Jet keeps processing data without loss when a node fails, using a replicated in-memory data store. You can add nodes to a live cluster, and they immediately start sharing the computation load.
Process Data from Multiple Systems
Process Data from Multiple Systems
Jet provides source and sink connectors for text files, Avro, JSON, Parquet, Apache Kafka, Apache Pulsar, ElasticSearch, relational databases including change data capture, Hadoop, Amazon S3, Azure Data Lake and many more. You can read from one and write to another, or aggregate and join the data in-flight.
Sub 10ms Latency at the 99.99th Percentile
Sub 10ms Latency at the 99.99th Percentile
Jet's core execution engine was designed for high throughput and low overhead and latency. In rigorous tests, it stayed within a 10-millisecond 99.99% latency ceiling for windowed stream aggregation. The engine uses coroutines that implement suspendable computation, allowing it to run hundreds of concurrent jobs on a fixed number of threads.
Production-Ready Out of the Box
Production-Ready Out of the Box
Jet nodes automatically discover each other to form a cluster, both in a cloud environment and on your laptop. It is lightweight enough to run on a Raspberry Pi. No ZooKeeper or Hadoop cluster required for production.
Use Jet To
Perform Streaming and Batch Analytics
Perform Streaming and Batch Analytics
Ingest data from a wide-variety of batch and streaming data sources, perform transforms and stateful computations on it, and write the results to the destination of choice. You can also cache the result set in-memory and serve it directly through thousands of concurrent low-latency queries and fine-grained, key-based access.
React To Real-Time Events
React To Real-Time Events
You can instantly react to real-time events with Jet, enriching and applying inference at scale. A single node is capable of windowing and aggregating 100Hz sensor data from 100,000 devices with latencies below 10 milliseconds: that's 10 million events/second. Jet works with many streaming data sources such as Apache Kafka, Apache Pulsar, or message brokers such as RabbitMQ.
Build Stateful Workflows
Build Stateful Workflows
Use Jet to build distributed and stateful workflows. Ingest data, denormalize and process it, run a series of distributed computations and cache the intermediate results in queryable memory and finally write the results to your destination of choice.
Who is Using Jet?
