Jay Mason, Associate Partner, Big Data & IoT M&S Consulting Industrial IoT Platform Services for Analytics and Machine Learning [843708] Jay Mason, Associate Partner, Big Data & IoT M&S Consulting Oracle BIWA Summit 2017
Speaker Bio - Jay Mason 20+ years working with Oracle Business Intelligence and reporting solutions Worked with machine data (SCADA PLCs) since the 1980s Completed multiple successful IoT implementations from POC to Production Leader of Big Data, IoT and IDM practices at M&S Consulting jay.mason@mandsconsulting.com
Agenda Overview of IoT IoT Use Cases in Elite Sports Industrial IoT Platforms Industrial IoT Analytics Industrial IoT Deployment
How This IoT Presentation is Different No predictions on the size of the IoT market No Gartner Hype Cycle
Challenges of Industrial IoT
IoT Analytics Predictions for 2017 New Predictions for 2017 The “edge” will become a huge growth market DDoS attacks from IoT devices will continue AI/Deep Learning will become a reality And the common predictions from prior years will be repeated Self-Service BI will become the new normal More disparate data sources and formats More actionable, automated decision capability
USA Cycling IoT Use Case Times automatically fed to IoT Platform using MQTT protocol Integration with other data feeds (power, heart rate, etc.) for advanced analytics Coaches and athletes have information right after each practice run
USA Cycling IoT Solution Architecture
US Speedskating Use Case Re-used solution from USA Cycling with minimal modification Challenge: Frequent connectivity issues and power outages Recommendation: Store copy of original messages for validation when proving out solution
Common IoT Platform Capabilities Security Analytics Hadoop scale data management and storage (Near) Real-time data insights Application Development & Integration HTTP, MQTT, CoAP SDKs / APIs Device Features Device commands, provisioning, device registries, libraries, and virtual devices Workflow / Event Processing Monetization / Marketplace
End-to-End IoT Security Unique device identity Authentication Transport-level security Device metadata and lifecycle management Secure every layer from the device to the cloud
Industrial IoT Platform Comparison mandsconsulting.com/industrial-iot-platform-comparison
Sample Industrial IoT Platforms Amazon AT&T Autodesk Ayla Networks Blackberry Bosch Bright Wolf Strandz C2M C3 IoT Carriots Cisco Jasper Concirrus Connecthings Connio Cumulocity Davra Networks Device Insight Echelon Emerson EVRYTHING Exosite FIWARE GE GroveStreams HP IBM Intel Itron Kaa Losant M2Mi MachineShop Microsoft Octoblu Oracle PLAT.ONE Preva Systems PTC Salesforce SAP Sentilo Siemens T-Mobile thethings.iO ThingPlus Verizon Waylay Wind River Workpad IoT Wot.io Xively Zebra Technologies
IoT Capabilities - Connect, Analyze, Integrate Source: Extending Enterprise to the Edge | White Paper | Oracle and Wind River
IoT Platforms and Spark Integration Amazon IoT and Apache Spark on Amazon EMR IBM Watson IoT Platform and IBM Analytics for Apache Spark for Bluemix or IBM Streaming Analytics for Bluemix Intel IoT Platform and Intel Data Analytics Acceleration Library (Intel DAAL) Microsoft Azure IoT Suite and Apache Spark for Azure HDInsight Oracle IoT Cloud Service and Oracle Big Data Cloud Service SAP HANA Cloud Platform for IoT and SAP HANA Vora
IoT Platforms with Integrated ML Amazon IoT and Amazon Machine Learning GE Predix IBM Watson IoT and IBM Watson Machine Learning (Coming Feb. 15th) Intel IoT Platform Kaa IoT Platform Microsoft Azure IoT Suite (includes Azure ML) SAP HANA Cloud Platform for IoT and SAP HANA Smart Data Streaming thethings.iO ThingWorx IoT Platform Verizon ThingSpace IoT Platform
IoT PaaS Marketplaces AWS Marketplace Predix Catalog IBM Cloud Marketplace Microsoft Azure Marketplace Oracle Cloud Marketplace ThingWorx Marketplace Salesforce AppExchange SAP HANA App Center
Oracle IoT Cloud Services Analytics Select raw data streams Choose a data analysis pattern Route analyzed streams to services/applications
With IoT, Context is King Context defines the state of an environment (usually the user’s environment) in a certain place at a certain time.
IoT Integration and Workflow (CBM = Condition Based Maintenance; PdM = Predictive Maintenance)
Oracle IoT Cloud Service Integration Oracle Business Intelligence Cloud Service Oracle Integration Cloud Service Oracle Mobile Cloud Service Enterprise Applications Oracle E-Business Suite Oracle Asset Tracking Oracle JD Edwards EnterpriseOne Oracle Transportation Management
Reference IoT Layered Architecture
Industrial IoT Analytics Architecture
Centralized IoT Data Store Store data streams from many disparate data sources Standardized data regardless of device protocols Drives analytics for realizing the benefits of IoT Centralized repositories are where big data and analytics can be applied to extract insights (Preferably) Optimized for time series analysis
IoT Data Processing Options
Edge Computing/Analytics Move the processing to where the action is Minimize data transfer Minimize network latency
Cloud and Fog
Digital Twins Oracle IoT Cloud, GE Predix and other major IoT PaaS services offer implementation of the Digital Twin through: Virtual Twin Predictive Twin Twin Projections
Twin Projection Purpose: Integration of insights Triggering appropriate processes/workflows Accessing data for decision support from business apps Allowing business apps visibility into current and predicted machine states and environment
Predictive Modeling with Oracle IoT Cloud Service Spectrum of complexity Simple models typically based on trends and patterns For these, Oracle Stream Explorer (included in the IoT Cloud) is sufficient More complex models can be created using Apache Spark based analytics engine in the Oracle IoT Cloud Typically developed using Oracle R Advanced Analytics for Hadoop (ORAAH) These R models can then be executed in the IoT data pipeline Oracle IoT Cloud offers libraries for working with Time Series data Business users can use simpler interfaces provided by the Oracle Big Data Discovery product
Machine Learning Integration Machine Learning capabilities integrated with Industrial IoT Platforms Built-in data processors Training models Automation Alerts / Monitoring Tuning Visualization tools
IDC MaturityScape IoT Maturity Model
IoT Phased Deployment from IDC Connecting devices and assets Applying real-time and predictive analytics and machine learning Achieving service excellence
Key Takeaways Connect, Analyze, Integrate Edge computing/analytics minimizes data transfer and latency Context is King Device data does not exist on an island → Integrate!