Dan’s Data Notes — Weekend Thoughts On Edge Computing
There are many trends in technologies developing in front of us. One one of the most fascinating ones to me is Edge Computing given its implications for Data Analytics and potentially Data Security.
What is Edge Computing?
One of the simplest definitions I could find is that Edge computing refers to having the ability to move computations as close to the data as possible. This prevents having to send data back to a central processing server which can have some implications for the speed of the results as well as data privacy and security implications.
The general idea is that the data is processed in real-time at the source and any data filtering, caching and aggregations are also done at the source. This distributed processing paradigm also has some architecture challenges.
Why is it possible now?
IoT: 1) Data is being generated by an increasing number of sensors and devices that become online 2) Sensor cost and the device components continues to decrease 3) The processing power of the devices continues to increase as more hardware fits in smaller devices
5G: 1) Higher throughput than traditional networks 2) Allows for an increased number of connected devices 3) offers lower latency 5) Helps extend power consumption in mobile and lightweight devices.
Analytics Technology: 1) Analytics models require a lot of resources to train the models but once trained the inference cost is very low 2) Lightweight database technologies have been developed 3) Specialized lightweight machine learning libraries have been developed.
What are some key implications?
Application User Experience: True real-time low latency use cases are now possible without specialized devices.
Analytics Capabilities: Given the bulk of machine learning model training can be done in advance, edge devices can perform fast inference operations for most machine learning models.
Data Security: Since the data never leaves the device or at least no sensitive data, it helps to better prevent data breaches.
