Get started Bring yourself up to speed with our introductory content.

Cloud versus fog computing for IoT: The fog is rolling in

The relationship between the cloud and IoT is evolving. Today’s IoT technologies integrate a mix of processing and analyzing data at the edge, in the cloud or nearer the ground — in the “fog.”

But what does fog computing for IoT have to offer?

Cloudy with a chance of …

IoT typically involves the collection and analysis of massive amounts of data. How much data? Cisco estimates that by the end of 2019, IoT will generate more than 500 zettabytes (500 trillion gigabytes) per year in data. A lot of that data will be sent to the cloud to be processed, examined, analyzed, digested and ultimately acted upon. Cloud computing allows companies to avoid or minimize upfront IT infrastructure costs and lets them focus on their core businesses instead of spending resources on computer infrastructure and maintenance. Cloud also has the big benefit of immediate availability of almost unlimited capacity. And, the cloud has become the go-to solution for use cases needing business analytics.

Of course, putting all your data on someone else’s computer, while cost-effective, does pose some risks. As with any system, there are technical and security risks associated with the cloud. Downtime caused by technical outages is inevitable and occurs when cloud service providers become overwhelmed in the process of serving their clients. According to the Cloud Security Alliance, the top three technical causes of security risks in the cloud are:

  • Insecure interfaces and API’s — 29%
  • Data loss and leakage — 25%
  • Hardware failure — 10%

On the other hand, security in the cloud is often as good as or better than other traditional systems, in part because service providers are able to devote resources to solving security issues that many customers cannot afford to tackle or which they lack the technical skills to address. However, when it comes to cloud and IoT, the complexity of security is greatly increased when data is distributed over a wider area or over a greater number of devices, as well as in multi-tenant systems shared by unrelated users.

The greatest risk for hosting IoT systems purely in the cloud, though, is the business risk. Sending data to the cloud, performing analytics and sending back actions to implement is not an efficient system for microdata transactions that are incredibly latency sensitive.

Living on the edge …

In many real-world use cases, some of the data sources in an IoT system may not be continuously connected to a network, such as with autonomous vehicles, implanted medical devices, fields of highly distributed sensors used in geofencing of gas or oil pipelines or water quality monitoring systems in regional dams, rivers, lakes and oceans, as well as underground water reserves, and mobile devices. Some data analysis needs low latency, including maintenance alarms, conditioned-based monitoring and machine learning. Additionally, some data is personal and should be private to the personal device on which it is collected. These data and processes are better located at or near the edge, where the internet connects with the physical world or end users.

What is fog computing?

Fog computing is a resource model that lives closer to the “ground,” at or near the edge, between smart end-devices and traditional cloud computing. Fog computing implies a distribution of the communication, computation, and storage resources and services on, or close to, devices and systems in the control of end-users. Fog computing is best seen as a complement to cloud computing, rather than a substitute as it can accomplish short-term analytics at the edge, freeing cloud resources to take on large data set’s tasks. At the same time, insights obtained by the cloud can help update and tweak policies and functionality at the fog layer.

So, are we talking about hybrid cloud?

In a word, no. In my opinion, the term “hybrid cloud,” was an invention of creative hardware vendors to market their on-premises products in the new “hip” way. The “hybrid” in the hybrid cloud is a mix of on-premises, private cloud (aka internal or corporate cloud) and third-party, public cloud services. A common hybrid cloud use case is “cloud bursting” — a situation where workloads are “spilled over” from private to public cloud environments to meet capacity demands, for instance, to handle a spike due to seasonal traffic or a news event. Hybrid cloud works well for disaster recovery situations, keeping a production environment in a private cloud and a recovery environment in a public cloud.

This fog won’t burn off

IoT is going to continue to connect more of everything. These connected deployments will require some combination of low latency, mobility, geographic distribution, network bandwidth, reliability, security and privacy challenges that preclude cloud-only systems. They will also require the kind of energy, space, capacity, environmental, reliability, modularity and security challenges that preclude fog-only technologies. And, in some cases, cloud solutions and fog solutions will, in fact, compete with each other. There will be new vendors and players that we would have called on-premises before that are doing compute-work that previously would have been relegated to the cloud.

So, for IoT systems, is it a cloud-versus-fog-computing decision? It’s not either/or — it’s both/and. When data analysis is time-sensitive, or you want purpose-built systems with unique needs, you can be better served by edge computing and/or fog computing, not as replacements for the cloud, but as necessary components in your IoT mix.

Fog computing is creating new opportunities and capabilities for industrial IoT deployments. Fog computing is also creating one more architecture decision to make. What do I keep within my four walls? What do I run in the unlimited capacity cloud and what do I move close to where to action is — into the fog? Don’t leave these decisions to the last minute. Instead, start small with ALL of these components in mind. Secure and test them early, look at reference architectures with this new ability in mind.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

Data Center
Data Management