Figure 1 : IoT Standard Design Pattern with Google Dataflow and Pub / Sub
cific application components for more seamless application changes over time .
Data Ingest Scalability
Successful IoT platforms also require the flexibility to ingest data from a continuously changing and growing universe of devices . Google Dataflow enables this capability by providing flexible interfaces for consumption of data from a variety of devices and sources . Native integration with the Google Cloud Platform ’ s compute and storage provide ample scalability for nearly any IoT application .
Dataflow enables the reliable transportation of data and in-flight data execution of work . Dataflow provides connectors for native Google services and enables developers to create custom modules for activities such as ETL , or predictive model execution . Dataflow can also be used to ensure business processes are implemented as consistently executed workflows .
Data Synchronization
Many IoT consumers will require the ability to utilize devices or applications offline or in a disconnected state . In those environments , data must be synchronized between local devices and a centralized data repository . Google Firebase [ 3 ] natively enables this process in the Google Cloud . Google Firebase allows applications to operate in an offline mode by synchronizing the data to a central location for later use and analysis .
Model Management
Model management ensures that deployed analytical models are properly tested , tracked and continuously updated in a consistent way . As analytic execution becomes more distributed , such as with the Analytics @ The Edge model , where model management becomes more critical due to the distributed nature of execution and the large number of devices with locally stored analytical models . IoT platforms using an Analytics @ The Edge design pattern leverage the computing resources within connected devices for the execution of predictive models and recommendation engines .
Successfully architected IoT platforms enable transparent elasticity . Google Pub / Sub and Google Dataflow allow software and operations teams to replace individual components of an IoT platform , minimizing downtime and increasing performance at the users ’ demand . These technologies allow IoT platforms to be built quickly and adjusted over time as the technology evolves and user demand increases .
SPRING 2016 | THE DOPPLER | 27