At the Data Science for Internet of Things course, we have been working with the Predix APIs. We emphasise the Industrial IoT and Predix from GE is one the best examples of a mature IIoT platform. Predix is currently available on a free trial and I asked GE about the process after the free trial. After 6 months of trail, developers would have to pay for the price listed under each service for example HERE I have used the documentation from the Predix site for this article.
What is Predix Platform?
The Predix platform is a cloud-based Platform-as-a-Service (PaaS) for the Industrial Internet. Predix provides a complete environment to create solutions to run industrial-scale analytics for Industrial IoT. Predix platform uses a Cloud Foundry based microservice architecture which enables it to deliver functionality as a set of very small, granular, independent collaborating services.
The high-level architecture of Predix is as below
The Predix platform provides the following industrial microservices:
Asset Service: to support asset modeling. Application developers use the Asset service to create, update, and store asset model data that defines asset properties as well as relationships between assets and other modeling elements. The Asset service consists of an API layer, a query engine, and a graph database. For example, an application developer can create an asset model describing the logical component structure of all pumps in an organization, then create instances of the model to represent each pump in an organization.
Analytics Services: can be used to manage the execution of analytics through configuration, abstraction, and extensible modules.
Data Services: include the following services: Time Series Service and the BLOB Store Service(Binary Large Object Storage). Data Services also include a SQL Database, the PostgreSQL object-relational database management system, Message Queue, Key-Value Store
Security Services: consists of User Account and Authentication, Access Control and Tenant Management
Machine Services: a device-independent software stack that allows you to develop solutions to connect machines to the Industrial Internet.
Predix Mobile: used for mobility applications
Event hub: Predix also includes the Event Hub beta pub/sub service built to be secure, massively scalable, fault tolerant, and language-agnostic. It supports textual (JSON over websocket) and binary (protobuf over gRPC) for publishing, and binary (protobuf over gRPC) for subscribing. Devices communicate with the Event Hub service by publishing messages to topics.Event Hub provides the ability to ingest streaming data from anywhere to the cloud for processing.
We are more interested in the Analytics service considering our emphasis on Data Science for IoT. An ‘Analytic’ is roughly equivalent to a model in traditional data science parlance. The Analytics Catalog service provides a software catalog for sharing reusable analytics across development teams. The Analytics Catalog service supports analytics written in Java, Matlab and Python. The Analytics Runtime service supports analytics from the Analytics Catalog service.
Developer Workflow for analytics is as follows
Stage 1: Develop the Analytic
Stage 2: Publish the Analytic to the Analytics Catalog
Stage 3: Validate the Analytic
(3a) Test the analytic with a sample data set and verify the results using the Analytics Catalog service.
(3b) If necessary, modify the analytic and upload new artifact to the catalog using the Analytics Catalog service.
Stage 4: Release the Analytic
Stage 5: Execute the Analytic
Predix is interesting for our work in the Data Science for Internet of Things course, because of its emphasis on the Industrial Internet. In subsequent articles, I will cover Predix Edge Processing capabilities and the Predix APIs. I am interested in finding an equivalent for Industrie 4.0. If you know such as service for Industrie 4.0 or you wish to join the course – please email me at info at futuretext.com
In next versions of this series, I will cover Predix Edge Processing and the Predix API
Predix is also covered in our forthcoming book on Data Science for Internet of Things by Ajit Jaokar and Jean Jacques Bernard