As companies take advantage of emerging IoT use cases, they face connectivity, data volume, and security challenges that affect network and system architecture. Choosing the right architecture at the lowest cost depends on a combination of factors, including:
- The number of IoT sensors and devices deployed for data collection
- The geographic distribution of IoT devices
- Whether IoT devices are mobile or stationary
- The volumes of data that the devices generate
- Whether the data stream is continuous or intermittent
- Whether real or near real-time data processing and analysis are required
- Regulatory and security requirements
- The available budget
When it comes to IoT, there is no one-size-fits-all approach. The optimal architecture varies depending on use case. What kind of architecture is right for your solution? In general, IoT architectures can be broken down into three categories: two-tier, three-tier, and four-tier.
Two-tier
Two-tier is the simplest, least costly IoT architecture, consisting of a tier of IoT devices generating data, which is then processed and analyzed in a public or private cloud tier. This works well for small deployments of up to a few hundred IoT devices in one or a few nearby locations. In most cases the devices are stationary, and real- or near-real time analysis is not required.
Two-tier is a perfect, low-cost solution for small retail IoT projects with one or two locations. It’s also a smart solution for environmental monitoring, which relies on a small number of local temperature, vibration, humidity, or light sensors.
Three-tier
Three-tier is the most common IoT architecture for several hundred or more IoT devices and real-time or near real-time analysis and response. It’s called three-tier because it adds a third layer of data storage and processing at the network edge, right next to the data source, for custom functions, application logic, and rule sets.
Four-tier
Four-tier works well for the largest mobile and/or globally dispersed IoT deployments, with thousands of devices. This architecture adds a fourth tier, often called the “fog computing layer,” somewhere between the edge devices and cloud data center. The fog tier may deploy one or several dispersed fog nodes to aggregate data from a few regional locations and run sophisticated data analytics as close to the edge as possible.
Once you’ve chosen an IoT architecture, selecting the best cloud solution means balancing budget, scalability, security, and control considerations. A private cloud may provide security, control, and customization beyond what is available in the public cloud. However, a private cloud may not offer the public cloud’s high level of elasticity, scalability, agility, flexibility, and low capital expense.
The public cloud is almost infinitely elastic and has little to no up-front cost, as most are structured in a way that allows businesses to pay for the space that they use only if and when they use it. The public cloud’s easy provisioning and scalability also allow the rapid deployment and scale of new IoT functions.
For many businesses, a public/private hybrid cloud architecture strikes the perfect balance of control, security, low capital cost, elasticity, and workloads matched to the right infrastructure. A carefully designed hybrid cloud architecture works well for companies seeking control without having to invest the huge sums in infrastructure that are only occasionally needed for peak loads. The final decision on combination depends entirely on the industry, business, solution, and goal.
When considering which type of cloud for IoT device deployment, it may be worth considering serverless computing.
Serverless Computing
Serverless computing vastly simplifies the operational aspects of IoT application development and deployment. Instead of calculating and provisioning the best level of server or container resources, the cloud customer simply takes advantage of automated platform as a service (PaaS) provisioning. Additional resources are added in real-time as needed.
Rather than explicitly managing provisioning and operations, PaaS users are hidden from the underlying infrastructure and pay based on actual resource consumption.
Serverless is gaining adoption in the cloud but also makes sense at the edge and fog layers. It’s a great solution for three- and four-tier architectures with many dispersed IoT devices generating huge amounts of data that require quick analysis. Fig. 6 shows a four- tier deployment model on a hybrid cloud with serverless computing capabilities in the fog and edge tiers.
Engage with IoT experience
Like any emerging technology, IoT has been evolving rapidly. By choosing the right IoT architecture and leveraging the flexibility, agility, and elasticity of the cloud along with groundbreaking technology such as serverless computing, companies can cost-effectively harness IoT for competitive advantage today while having the agility to adjust to what IoT will become tomorrow.