A little bit about Edge
Prangya Mishra
Associate Vice President - IT & Digital Solutions at JSW Steel | Head-MES | APS | IIoT Architect | ML, AI at Edge | Ex- Accenture, Schneider Electric, Wipro, Alvarez & Marsal | Metals SME | Creator of "Process In a Box"
Edge computing is not replacement of MES and related application and systems meant to digitalise the manufacturing. Edge computing is designed to extend the power of cloud to on-premise servers. You can use edge systems to connect your IT and OT systems and run Analytics to derive insights irrespective of “internet” connectivity.
Consider edge as an wrapper application around your existing IT / OT application constantly interchanging data with them and providing you with required insights to run your manufacturing efficiently. All these without having to be connected to the internet all the time
To generate the insights you need to build some intelligence and the intelligence is not thing but a set of complex rules and algorithms.
The first thing you do is define the use cases or what insights we would like to derive for efficiently managing your manufacturing. Once that’s done, you define the data points you will need in order to bring the use cases to live. Once the data points are finalised, you send the data to cloud and accumulate data for certain periods of time. Using that historical data you build prediction models of varying complexities. Once those models are built in cloud and validated by business users it is ready for being deployed in production shop floor usage. Note till now the prediction model is in cloud and in order to access any insights from them you need “internet” to access them.
Since you don’t want internet connectivity and latency ( and security ) to be a bottleneck in the process of automated predictive insights generation based on your shopfloor data, you now thought of putting the predictive insight generation mechanism a computer or server seating close to machine and not requiring a regular internet connection to be able to generate insights for the shop floor. This server is called the “edge server”
Next step would be how can you use the business validated predictive models residing on cloud on the edge server. There are mechanism available within cloud where the models can be pushed to the edge server just like the mobile OS updates we receive. Once updated the edge server now can work offline completely independent of availability of the internet.
You can decide which use cases you want to run on edge server and can push only those use cases from cloud to edge server.
The edge server can securely access data from all on premise IT / OT system including the sensor “locally” and based on use cases can start providing insights. Just connect the edge server to Internet only when you want to update some use cases or add new use cases just like upgrading our mobile apps.
I love the simplicity of edge and the flexibility it brings to the day to day operational Analytics and truly believe, Edge can transform manufacturing the way PLCs did in third industrial revolution.
[The views expressed in this blog is author's own views and it does not necessarily reflect the views of his employer, Wipro Limited ]
Automation | Industrial Software | Digital Transformation
4 年Sorry Cloud, the industry still need you, just away from operation critical application. I would be curious to look at the numbers for manufacturers Internet SLA who choose to go “all in” in the cloud with analytics solutions connected to their production and decision systems... and the related cost of unplanned delay.