It’s that time of year again, when everyone is unleashing their “Top-10” predictions lists on the world. As with years past, I’m only going to unleash on the world just one, solitary prediction for looking ahead.
For 2018, expect to see cloud computing morph into a new, more distributed form, which many observers and some vendors (mainly Cisco) are referring to as “fog computing.” I don’t want to fog up the works with new terminology, so let’s just cut to the chase and clear the air on what is happening. Essentially, there’s a movement away from big centralized clouds, to more highly distributed clouds that are closer to their data origins and help cut down on latency.
The US office of National Institute of Standards and Technology (NIST) has weighed in on fog computing with a document of its own, so it can be said that it officially is now a thing. NIST defines fog as “a horizontal, physical or virtual resource paradigm that resides between smart end-devices and traditional cloud or data centers. This paradigm supports vertically-isolated, latency-sensitive applications by providing ubiquitous, scalable, layered, federated, and distributed computing, storage, and network connectivity.”
Fog brings processing closer to the edges. Fog can mean anywhere along the continuum between the central cloud and outlying edge devices, be they senors, applications or mobile handhelds. For example, as the NIST paper points out, a fog architecture may consist of "large-scale sensor networks to monitor the environment." The electrical Smart Grid is another example of a fog architecture, "requiring distributed computing and storage resources."
Fog computing is going to be taking off in 2018 in a big way. A study conducted 451 Research, commissioned by the OpenFog Consortium predicts the fog computing market is projected to exceed $18 billion by 2022, growing 18-fold in size from a base of $1.032 billion in 2018.
There’s another trend fueling fog – the rise of micro data centers. There is so much capacity that now can be packed into small, densely packed processors that it makes economic sense to deploy networks of such devices as powerful nodes. As Scott Fulton III recently explained in a ZDNet post, these will be the small engines that could powering fog computing architectures – by harnessing processing and analytical power where it is needed, reducing the potential network latency that more centralized cloud architectures incur. Thousands of very small data centers, positioned near cell phone towers to collectively supply as much compute power as one of Amazon’s sprawling complexes. These small data centers may offer a worthy alternative to the large cloud-based services. In his article, Fulton quotes Egal Elbaz, AT&T vice president for ecosystem and innovation, who says this is feasible due to "the evolution into 5G, ultra-low latency and high throughput," along with "an ability to orchestrate in real-time, based on demand and consumption."
The NIST document lays out the main characteristics of fog versus traditional, more centralized cloud as we've known it over the years:
Contextual location awareness, and low latency. "Because Fog nodes tend to sit very close to the IoT endpoints, analysis and response to data generated by the endpoints is much quicker than from a centralized cloud."
Geographically distributed. "In sharp contrast to the more centralized cloud, the services and applications targeted by the fog demand widely distributed deployments. For instance, the fog will play an active role in delivering high quality streaming services to moving vehicles, through proxies and access points positioned along highways and tracks."
Real-time interactions. "Important fog applications involve real-time interactions rather than batch processing."
Predominance of wireless access. "Although fog computing is used in wired environments, the large scale of wireless sensors in IoT demand distributed analytics and compute. For this reason, fog is very well suited to wireless IoT access networks."
Heterogeneity. Fog nodes come in different form factors, and will be deployed in a wide variety of environments, and the devices they collect data from may also vary in form factor and network communication capability."
Interoperability and federation. "Seamless support of certain services (real-time streaming services is a good example) requires the cooperation of different providers. Hence, fog components must be able to interoperate, and services must be federated across domains."
The NIST report also maintains that there will be a great deal of interplay between the fog and the cloud. "The fog is positioned to play a significant role in the ingestion and processing of the data close to the source as it is being produced. While fog nodes provide localization, therefore enabling low latency and context awareness, the Cloud provides global centralization. Many applications require both fog localization and cloud globalization, particularly for analytics and big data. Fog is particularly well suited to real-time streaming analytics as opposed to historical, big data batch analytics that is normally carried out in a data center."
Previous half-baked predictions:
2015: Cloud computing becomes just "computing"
2016: Cloud becomes corporate innovation machine
2017: The rise of the Internet of Things
(Disclosure: I am also a contributor to ZDNet, mentioned in this post.)
