Unlike the next wave of cellular networking technology (5G), the next wave of edge computing is officially here. While 5G works through fixing vulnerabilities for widespread use, edge computing services and hardware are currently being delivered at scale. And there’s even more to come…
By the end of 2020, Vapor IO plans to build 20 edge data centers. The first one is now live in Chicago and, during a test, completed a facial recognition job in 13 milliseconds compared to the average 240 milliseconds. Also, SmartNIC cards that are already in production are being improved to further reduce processing time.
These topics and more are covered in this fifth issue of High Performance Infrastructure!
Featured Stories
About: Sensors. Data processing. New hardware being developed for edge compute. Chip comparisons.
- There is widespread agreement that moving massive amounts of data to the cloud for processing in bulk is unworkable from a power, bandwidth and latency perspective.
- 5G will not alleviate the need for an edge layer due to the limitations of millimeter wave technology and the time and energy necessary to move that data.
- More data is better for training systems, but inferencing can operate on a much smaller data set, making it suitable for the edge.
- Smart card tech is being developed for the edge to handle data processing from cards rather than in a CPU or GPU.
About: Security flaws in 5G that support StackPath’s CEO’s claim that widespread use of 5G Is still 24-36 months out.
- At the Association for Computing Machinery’s Conference, researchers presented new findings that the 5G specification still has vulnerabilities.
- 11 vulnerabilities were found that could lead to exposing user locations, downgrading service, running up wireless bills, and tracking when users make calls, text, or browse the web.
- Since one of the purposes of 5G is to hide a user’s location, researchers are pressing GSMA, the standards body for mobile operators, to make fixes.
- Researchers identified vulnerabilities with a new tool called 5GReasoner.
About: Developing better hardware and tech to meet the demands of edge computing.
- Intel believes that in five years we will create 10 times more data than we did this year.
- 70% of that data will be created at the edge and only half will make it back to a public cloud. The rest will be stored, processed and analyzed at the edge, which requires a whole different approach to hardware systems.
- GPU is great at some things, but not everything. Just like you wouldn’t go running in leather shoes, you’re not going to throw a GPU and an inference workload at the edge.
Other worthy reads
