AT & T Labs, the United States, has been quietly defining its concept of 'edge computers' and is slowly pushing deployment, which in the long run has a far-reaching impact on the design of cloud and mobile systems in the future. Computing is defined as a clustering of servers and storage systems placed around its network in order to provide low-latency services; the company foresees the widespread presence of such systems that vary in size and location, depending on the applications and needs.
Alicia Abella, senior director of AT & T Labs, said: 'The edge of the operation is the next step for us to make the network full of performance, and we are busy building the edge of the computing architecture;' she is in the Fog World Congress conference in Silicon Valley, "We want to deploy the edge operations node in the mobile data center at the building, the customer's location, and the headquarters office, depending on where and where our spectrum is, and we are developing a way to optimize the location . '
The edge system is available for a variety of purposes, with the goal of assisting AT & T to reduce the amount of data that the core network can carry and deliver; it also provides higher quality for existing services and is expected to spawn new services. One of the significant applications Is a video analysis for the preservation of the camera, which may use a graphics processor (GPU), an FPGA, or other accelerator solution and is deployed in the city.
A more challenging application is the task of supporting autopilot vehicles because such applications require substantial investment in roadside infrastructure and are currently uncertain of their investment returns; it is noteworthy that Abella revealed that AT & T's current There are 12 million smart cars on the network, and the number is growing at a rate of one million per quarter.
AT & T says smart vehicles are one of the use cases that edge computing can support (source: AT & T)
Seeking small and medium-sized GPU accelerator program
Different edge systems will be tethered to the AT & T core network by different parties, some may use 1 or 2 compute racks, others may be larger, and Abella said the company has so far faced the challenge of hardware It is to seek medium and small versions of GPU accelerators, because such programs are usually large size, high power version.
An early prototype for AT & T was putting the camera in the car, controlling it with the Raspberry Pi development board, and sending the video to a remote monitor; Abella said: 'The size of the edge computing system may be as small as possible The actual situation is that we foresee some of them as a walk-in closet ... We are now only beginning to think about the tactics of deploying this system. "
AT & T did not provide technical details or a timeline for its edge computing system architecture (source: AT & T)
Once the system is widely deployed, it can bring a significant impact on the mobile system; mobile devices can be part of the task by the edge of the computing system to share, save battery power; Abella explained: 'imagine a new generation of mobile devices, or not Need to be bound with the computer, the use of mobile edge operation of the AR / VR equipment ... there are ways to get signal processing at the edge of the way.
AT & T's ultimate goal is to define a software platform for edge computing, providing developers to write applications; but the current concept has just formed, may be achieved until 2020. Abella is expected in the distant future, this platform will Will be similar to SETI @ home that use the home PC search for extraterrestrial intelligence creatures 'crowdsourcing of compute' project, participants can get such as bitcoin (bitcoin) such remuneration.
Fog World Congress brings together professionals from different fields such as edge computing, computing, and cloud computing. At the venue, some vendors have demonstrated miniaturization, rugged gateways and industrial servers for edge computing applications The
Compiler: Judith Cheng