I read an interesting article on Fog Computing and thought readers might like a short precis:
Applications such as health-monitoring or emergency response require near-instantaneous response such that the delay caused by contacting and receiving data from a cloud data-centre can be highly problematic. Fog Computing is a response to this challenge. The basic idea is to shift some of the computing from the data-centre to devices which are closer to the edge of the network – so moving the cloud to the ground (hence “fog computing”). The computing work is shared between the data-centre and various local IoT devices (e.g. a local router or smart-gateway).
“Fog computing is a paradigm for managing a highly distributed and possibly virtualized environment that provides compute and network services between sensors and cloud data-centers” (Dastjerdi et al. 2016)
While cloud computing (using large data-centres) is perfect for analysis of Big Data “at rest” (i.e. analysing historical trends where large magnitudes of data are required and cheap processing necessary) fog computing may be much better for dynamic analysis of “data-in-motion” (data concerning immediate ongoing actions which require rapid analytical response). For example an Augmented Reality Application cannot wait for a distant data-centre to respond when a user’s head it turned. Similarly safety-critical and business-critical applications such as health-care remote monitoring, or remote diagnostics cannot rely on permanent availability of internet connections (as those in York know when floods knocked out their internet for days this year).
Privacy concerns are also relevant. By moving data-analysis to the edge of the network (e.g. a device or local mobile phone) which is often owned by, and controlled by, the data-source the user may have more control over their data. For example an exercise tracker might aggregate and process its GPS data and fitness data on a local mobile phone rather than automatically uploading it to a distant server. It might also undertake data-trimming so reducing the bandwidth and load on the cloud. This is particularly relevant as the number of connected devices increases to billions. This gain should be balanced with the challenge of managing an increasing number of devices which must be secured to hold sensitive data safely.
Another challenge is the climatic damage this new architecture poses. While data-centres are increasingly efficient in their processing, and often rely on clean-energy sources, moving computing to less efficient devices at the edge of the network might create a problem. We are effectively balancing latency with CO2 production.
For more information on see:
Dastjerdi, A. V., Gupta, H., Calheiros, R. N., Ghosh, S. K., and Buyya, R. 2016. “Fog Computing: Principles, Architectures, and Applications,” in Internet of Things: Principles and Paradigm. Elsevier / MKP. http://www.buyya.com/papers/FogComputing2016.pdf
One thought on “What is Fog Computing?”
Great WSJ article on this topic: