On 29/10/2021, the meetup of Edge Computing was held by FPT Technology Innovation (FTI) department with the presence of Nick Barcet – Senior Director of Technology Strategy in CTO office of Red Hat and Phan Thanh Son – member of Board of Management in FIS. There were some interesting information around the topic of Edge Computing i would like to sum up and share with all of you.
1. Definition of Edge computing
It is not a new concept. It has been mentioned for a long time when people would want to find a solution to have quick responses of processing data for use cases in which realtime, speed are critical factors. So, in a nutshell, edge computing can be easily known as moving computing resources (ram, cpu, storage, etc.) as near as possible to the sources of data. Sources of data are devices, means or anything where data is generated.
2. Misunderstanding of edge computing
It is common to misunderstand that some use cases of using IoT out there are edge computing, one of them is camera surveillance. Using specialized hardware for those IoT use cases is not purpose of edge computing. It makes some difficulties when working with specialized hardware including operating, maintaining and cost. Edge computing nowadays is focusing on commodity hardware where it makes comfortable to deploy new applications on edge, mitigate problems of day1 and day2 as mentioned in specialized hardware.
There should be a centralized management plane to manage multiple edge locations from day0 to day2. It is not reasonable to manually deploy, manage and maintain multiple edges since it will create more costs.
3. What are main technologies in edge computing
- Container: Lightweight, scalable, resilient with multiple management platform
- AI/ML: Trained models should be deployed at edge location
- Communication channel: Data should be securely transferred from edge device to edge back and forth, e.g. RF, Bluetooth, 5G
- Data pipeline: A pipeline to define which data should/shoudn’t be transferred from edge to cloud.
4. Security in edge computing
It is not only for edge but for all the IT technologies. In edge, from hardware to software running on edge, security must be considered carefully. All the data transferred from device to edge or edge to cloud must be encrypted as well as communication channel. There is a rising technology is Zero-trust architecture.
5. Cost when using edge computing.
Certainly, it will be more cost if using edges not in a right place. Adopting edges in the use cases where unnecessary will create more troubles in managing technologies, people. Edges will be useful in use cases where the autonomy is priority. For instance, autonomous car, IoT, industrial robot, etc. where data must be processed and decisions must be generated quick and precise based on processed data. In fact, all the vertical markets are discovering, exploring and also applying edge computing varying from Telco, Industry to Healthcare.
