IBM Streams

Continuously and rapidly analyzes massive volumes of moving data.

Can handle very high data throughput rates (millions of messages or events per second).

Provides sub-millisecond response times for real-time analysis.

Offers a specialized programming language (Streams Processing Language) for developing streaming applications.

Supports development in other languages like Java, Scala, and Python.

Provides IBM Streams Studio for simplified application development (for on-premises installations).

Executes applications on a single or distributed set of resources.

Can be installed and deployed on Red Hat OpenShift or Kubernetes environments using Kubernetes operators and Docker images.

Available as a service for IBM Cloud Pak for Data environments.

Automatically deploys stream processing applications on configured hardware.

Allows extending stream processing applications without restarting.

Provides security and auditing features for the runtime environment.

Continuously monitors the state and utilization of computing resources, with dynamic monitoring of running applications via Streams Console, streamtool commands, and REST APIs.

Capable of processing virtually any data, whether structured or unstructured (texts, audio, images, voice, video, web traffic, GPS data, email, financial transactions, satellite data, sensor logs).

Integrates with other data infrastructure like Hadoop, Spark, and external data sources via edge adapters. Can also integrate with Business Rules software like Operational Decision Manager (ODM).

Supports common stages in streaming applications:

Consumes live data from disparate sources (sensors, social media, databases, file systems).

Parses, transforms, filters, cleans, aggregates, or enriches data in memory for real-time analytics.

Performs real-time analysis, gaining insights using built-in toolkits (timeseries, R toolkit) or custom operators.

Creates logic to act on insights, potentially integrating with business rules engines.

Sends analysis results to visualization servers, sends alerts, or publishes results to subscribers.

Provides a set of toolkits for data analysis (e.g., timeseries for modeling, anomaly detection, forecasting; R toolkit for R scripts).

Designed to manage high availability, automated fault tolerance, and recovery.

Helps reduce data storage by filtering and extracting only relevant information in real-time.

Determines where best to deploy operators to meet resource requirements of new and existing workloads.