What is fog? What do you imagine when think about fog? Normally, when people mention fog, most of them usually think of London streets, the San Francisco Bay, or some of the famous places in Vietnam like: Da Lat hillsides, Ba Na mountain…
This interesting change is due to the invention of a computing model called Fog Computing. If the Cloud is a data center hosted by big providers like Amazon, Microsoft, IBM and Google, far away from the business’ data centers and computers, the Fog can be imagined closely to the human perspective as “the Cloud in which is nearer to the ground”. Fog computing is defined as a computing model where you collect and pre-processes data from sensors at a local gateway server before uploading it to the cloud for central processing and storage. It is not a revolution, many companies have done this from the beginning of time. The Fog can co-exist with The Cloud, and often acts as a middle layer between the sensor and the cloud. (Unlike a centralized cloud providers, a fog is local and distributed by nature). But now it has gained renew attention because Cisco has come into the game with their own Fog computing interpretation, along with their ecosystem, framework and tools.
The Fog and The Cloud
Imagine the case when you have to develop a face recognition application for public camera. Here we consider two designs:
DESIGN 1: The camera will send back all the images to the Cloud to be processed there. Actually, this has several disadvantages: It requires sending a lot of data through the Internet, which is rather slow. All the computations will be done at the cloud side, so the computation cost and data storage cost might be very high. The latency makes it unsuitable in cases where immediate action is required. Finally it requires that the camera hardware must support a stable internet connection to the cloud, which is unreasonable in this case.
DESIGN 2: The camera will send the images to a gateway device near the camera. Some computation, in this case, image pre-processing, segmentation and representation will be done at the Fog layer. Only after that the Fog will send back the partial result to the Cloud for further processing. This design doesn’t have the above design’s advantages. However, it has a drawback: You have to deploy and maintain some kind of gateway device.
In the age of Internet of things and big data, the amount of data that is generated by sensors is increasing dramatically. A modern racecar can generate gigabytes of data hourly through its 500 sensors. Let think about all the amount of data daily devices like lamps, refrigerators, smart cars, smart TVs… can push into the network if they are able to. And though it’s true that we are living in the age of 4G and ultra-high speed Internet, those will easily crumble under the weight of our data. So one of the solution to reduce the impact of big data to the network is through pre-processing data at the edge (or as near the edge as possible), and sending the result to the cloud for further and better processing.
In reality, we, FPT Software, have seen many implementations that can be mapped directly to DESIGN 2 as was mentioned in above. The gateway device can be a server, a personal computer, special self-manufactured devices, an iPad or a mobile phone, so on and so for. The questions we often see are “How can we build a gateway device?” “Is it cheap?” “Is it easy to use and maintain?” “Is there a framework for that gateway?” “Can we reuse the code across different gateway and devices and applications?” “How can it handle scalability and elasticity?” Fog computing gives us the answers (though may be an imperfect one) to those questions.
Quote from Cisco: “Fog computing is a paradigm that extends Cloud computing and services to the edge of the network. Similar to Cloud, Fog provides data, compute, storage, and application services to end-users. The distinguishing characteristics of Fog are its proximity to end-users, its dense geographical distribution, and its support for mobility. Services are hosted where they’re used: at the network edge or even end devices such as set-top-boxes or access points. By hosting services locally, the Fog paradigm reduces service latency and improves QoS, resulting in superior user-experience.” Cisco is the first player to promote a computing model for Fog Computing. In fact the word “Fog” may originate from their document. The original idea is from the Cisco ambition that wants to provide the programming framework for IoT applications to run on Cisco network infrastructure (Where Cisco routers may act as the gateway device). This has a lot of appeal for enterprise customers, with Cisco’s current domination in the enterprise scene. A preview of Cisco’s Fog Computing platform IOx is available athttps://developer.cisco.com/site/iox/ (Currently available to only a limited number of devices). The architecture is as below:
Yes, the future will be foggy. Fog Computing is currently stand strong with Cisco’s steady support, and Cisco itself is considering to widen the ecosystem with a large number of big and open source players. Fog Computing can leverage the current trend of Software Defined Network (SDN) both in the datacenter and the networking devices. Add in the IoT revolution, and it seems that Fog Computing’s future is assured. However, Fog Computing is still a young technology (like Cloud at its beginning) and we expect to see a lot of ups and downs. We certainly take a look on the potential of Fog Computing, but we don’t know which ecosystem, standards and protocol is winning. The next twelve months will be crucial to determine which player will dominate the Fog. Is it Cisco with its strength in network and edge? Is it AWS and Azure with its existing cloud base? Is it a player from the SDN open source movement? Or is it some kind of container solution that supports Fog Computing best?
About our plan, FPT should consider joining Cisco’s Fog Computing ecosystem. With our experience in cloud, embedded and IoT development, we can quickly learn and adopt Cisco technology. That will give us an early start and many opportunities to jump in a large, unoccupied IoT market, and get more close with a couple of our big customers, who now pay their cares and want to locate the position on IoT segmentations.
– VietPH –