Differentiation

---

For most businesses, the Internet of Things (IoT) brings three fundamental challenges:
(1) handling the tsunami of data coming from sensors and smart devices,
(2) detecting and responding to significant events as fast as possible, and
(3) providing an integrated view of historical and current business performance.
Meeting those challenges requires a scale-out Big Data approach to real-time event processing, and mechanisms for integrating data and analysis across batch and real-time domains.

Technologies such as Apache Storm provide a robust programming environment upon which data scientists and engineers can build real-time applications. However, those applications must be programmed, deployed, managed and integrated by hand. The amount of manual crafting required can come as a shock to organizations accustomed to the levels of abstraction and automation provided for years by traditional data analysis tools.

That’s where Dragonfly Data Factory™ comes in.

We’ve developed Dragonfly DataSwarm, a real-time IoT analytics and automation platform designed to:

Accelerate real-time application development

• Rapid GUI-based composition of applications – from data-source, analytics, sink and action components. Out-of-the-box components include real-time sources/protocols (Kafka, MQTT, AMQP, XMPP, Kinesis); ETL functions; analytics; and alerts.

• Component framework (SDK) – enables easy creation, management and reuse of additional components.

• Multiple stream analytics execution-logic formats – including SQL, Java, and Predictive Model Markup Language (PMML).

• Global online component library – constantly updated by Dragonfly and an ecosystem of developers.

• Local online component library – for sharing and reuse of components within an organization.

• Built-in analytics dashboard – with an SDK for creating new display components.

Integrate real-time and batch-based analytics

• Data integration – enable real-time use of batch data, and batch use of device-generated event data, including real-time query.

• Processing and data orchestration – update static repositories and/or trigger batch-job execution in response to real-time events (received or derived by analytics); dynamically update in-memory reference data used by real-time applications based on source-data change events.

• Integrated visualization – customizable displays, mixing real-time events and responses with historical data views, trends, inflection points, etc.

Simplify real-time application deployment lifecycles

• Application management across execution-engine clusters – Deployment, un-deployment, rebalancing, starting, stopping, monitoring, etc.

• Execution-engine cluster resource management – Add/remove nodes, performance monitoring, etc.

• Model-based, preemptive tuning and outage avoidance – monitoring and predictive analysis of real-time application and infrastructure performance.

• Non-disruptive updates – Adjust running-application component parameters, update running-application component code, and upgrade the DataSwarm platform without any application down-time.

• Multi-tenant support – Secure separation of platform users and their data.