Bench Talk

(Source: buffaloboy/Shutterstock.com)

In a 2016 report, Gartner predicted more than 20.6 billion Internet of Things (IoT) devices would be connected to the internet by 2020. That many devices bring new issues concerning connectivity, bandwidth, and storage. Here, we explore how the IoT evolves to meet these scaling challenges.

Scaling the IoT typically conjures up images of the cloud and the ability to elastically create new resources to meet the demand of a growing IoT network. The cloud is an important piece of the IoT, but there’s more to IoT scaling than just the cloud. Scaling is an end-to-end concept that touches the entire chain as well as the entire design, development, and deployment cycle. Let’s explore some of the key questions to consider when designing a large-scale IoT system.

The IoT is typically made up of three basic layers (although in practice, these layers can be considerably more complex).

• IoT Devices: IoT devices at the edge interact with the environment (device, person, appliance, etc.) and communicate to an IoT gateway.
• IoT Gateways: IoT Gateways serve as an aggregation point for the devices.
• Cloud: The data the devices communicate typically end up in a cloud, where the data are managed, analyzed, and presented to some interested party (a consumer, industrial application, etc.).

Now, let’s explore these layers and identify key questions to consider in the IoT design process.

IoT Devices

IoT devices are those objects in the field that produce data. They are the end products, where the gateways and cloud are necessary elements to sustain the IoT business.

As the front line to the IoT system, there are key questions to answer:

• How much data does a typical IoT device generate?
• How much data does a typical IoT device store?
• How much data does a typical IoT device consume (in the context of operating parameters, configuration, firmware updates, etc.)?
• What kind of security is required, and which requirements does this create (such as key servers)?

Next, let’s look at IoT gateways and the role they play in scaling.

IoT Gateways

IoT gateways mediate control and data between the device endpoints and the cloud. These gateways can be statically located (such as on a manufacturing floor) or dynamically located (such as in a roaming vehicle that scans water meters).

As the middle point of the IoT architecture, the IoT gateway must meet the requirements of the devices and the cloud, so its architecture is key to the success of an IoT system. The key questions to answer include:

• How many IoT devices will the gateway service?
• How much data in aggregate will the gateway service?
• Will the gateway process data (at the edge) in support of the cloud?
• Will the gateway compress or encrypt data (trading off processing for bandwidth)?
• Does the gateway support uniform access to devices, or are there peak times of traffic?

The next question: How does the cloud address scaling in the IoT?

The Cloud

The cloud is the orchestrator of the IoT system. It communicates with IoT gateways to receive data from IoT devices, communicates firmware updates and configuration information, and processes and analyzes data. If your IoT system is centralized, the cloud is the epicenter of data management and control.

The key questions to answer include:

• How much data processing is expected per device?
• What are the expectations for analytics or data science?
• How often are data collected from endpoint IoT devices?
• How often are endpoint IoT devices managed (firmware updates, monitoring, etc.)?

Other key questions for the cloud concern cloud architecture. Will your system rely on a public cloud such as Amazon, or will it use your own private cloud infrastructure? For more elastic and cost-conscious workloads, will a hybrid cloud infrastructure be used (where both private and public clouds cooperatively manage the IoT system)?

Many of the questions posed here illustrate the spectrum of scalability. At its root is the quantity of devices to be managed, but this management has a multiplicative factor. Scaling an IoT system goes beyond managing the data from IoT devices, but the scale of data processing and analytics is necessary for success, including manageability and security.

Whether you rely on elastic clouds for scaling of storage and processing or decentralize with IoT gateways to reduce cloud requirements, your end-to-end requirements must be carefully considered to ensure a balance. The management of IoT data and devices relies on a careful balance of processing, storage, and bandwidth for successful deployment.