The idea of connecting multitudes of smart devices to the internet goes back to the early ‘80s, when a modified Coke vending machine at Carnegie Mellon became the first internet-connected appliance (although it could only report on its inventory and whether the sodas were cold or not).

The term ‘Internet of Things’ didn’t emerge until 2000, but by 2010, the number of devices connected to the internet outnumbered the number of people connected.

Today, IoT is rapidly expanding across virtually all vertical industries and a dizzying array of device types. Over the last two years, adoption has been particularly strong in IT datacenter environments.


Adoption Status


Among 451 Alliance companies that have implemented or are looking at implementing IoT deployments, 43% are in production and another 25% are at the proof-of-concept/pilot project stage.

Adoption of IoT

More than 85% of the companies expect to be in the production phase of their IoT initiatives within the next year. Maybe... or maybe not.

Before you assess your peers’ progress on IoT projects – or benchmark your own internal progress – note that only about half (53%) of all IoT projects make it from the proof-of-concept stage to full production.

Moreover, the average IoT project takes about 18 months to go from conception to production due to both technical and business challenges – a long time to linger in pre-production purgatory.

That data comes from a survey of IT professionals in the 451 Alliance. The survey also provides information on key trends in the IoT space, including use cases, data collection/analysis strategies, the shift to network edge/perimeter processing and barriers to adoption of IoT.

The participating companies were split roughly evenly between those with >1,000 employees and those with <1,000 employees. About one-third have revenue of more than $1 billion, and two-thirds have revenue of less than $1 billion.

Highlighting the growing importance of the technology, spending on IoT is expected to increase 33% on average over the next year.


Report Highlights


IoT Moves to the Edge. Although most companies store and analyze IoT data in public clouds or in the corporate datacenter, there is a growing trend toward processing data at the network edge or perimeter.

Benefits Abound. Benefits of processing IoT data at the edge include improved security and performance, as well as potentially lower storage, connectivity and compute costs.

Hyperscalers Dominate. AWS and Microsoft are the most widely used IoT vendors among 451 Alliance companies, followed by Cisco, Google, Dell Technologies, Oracle, SAP, Hewlett Packard Enterprise and Siemens.


Endpoints and Use Cases


Organizations collect and analyze data from a wide variety of endpoints, ranging from retail point-of-sale equipment and environmental sensors to medical devices, robots and drones. But the most common endpoints reside in datacenter and IT equipment. In the context of IoT deployments, this includes a wide array of devices.

Datacenter Endpoints

The collected and analyzed data from these endpoints is used for an equally broad range of IoT use cases, most of which revolve around IT datacenter management, monitoring and automation.

IoT Use Cases in the Datacenter

Beyond the Datacenter


One of the thorniest decisions in all IoT implementations is where to execute the workloads. Security and cost considerations are the primary determinants, but IT managers also have to evaluate network connectivity, infrastructure resiliency, staff/expertise availability and latency requirements.

What Determines IoT Workload Execution Venue

Organizations have a wide variety of options for initial storage and analysis of the collected data but, according to the 451 Alliance survey participants, the top three are:

  • Within a public cloud (typically AWS, Microsoft Azure or Google Cloud)
  • In a remote datacenter environment (e.g., a server closet or micro datacenter) located close to where the data is generated/collected
  • In the enterprise’s centralized datacenter

On the Edge


According to the survey participants, these options are likely to remain the dominant ones for the foreseeable future, but there is a growing trend toward storing and analyzing data as close as possible to the devices generating the data; in other words, at the network edge or perimeter.

Options include:

  • On the endpoint itself
  • On a compute/network device attached to the endpoints
  • On an intelligent gateway device running analysis software

Among the 451 Alliance companies, 61% process at least some of the IoT data at the network edge/perimeter (24% on the data-generating devices and 37% on nearby IT infrastructure). The remaining 39% do not process data at the edge, typically sending it to a centralized IT datacenter or third-party service provider.

Thanks to superfast processors, advanced analytics software (including AI and machine learning) and edge-centric applications, more functions can be processed at the edge.

IoT Edge Processing Functions

There are a variety of potential benefits to processing data on the edge. Improved security is the #1 advantage, but increased performance (for both data analysis and data transmission, as well as reduced latency) and lower costs (storage, connectivity and compute) are also key drivers.

For these reasons, IoT edge processing is particularly advantageous for real-time analytics and applications that require immediate action.

Why Process IoT Data at the Edge?

Although many IT professionals cite improved security as a primary benefit of IoT edge processing, paradoxically, security is also one of the biggest concerns because processing data at the edge significantly expands the attack surface (which is true of IoT in general).

The types of data analysis performed at the edge are varied, but the most common are (in descending order):

  • Diagnostic analysis, filtering and monitoring (e.g., measuring health, performance and reliability of devices, systems and processes)
  • Real-time analysis of metrics and measurement data (e.g., sensor-based measurements such as temperature, speed, vibration, humidity)
  • A combination of traditional data analysis and cognitive computing and/or AI
  • Implementation of rules-based actions based on real-time data analysis
  • Data-intensive transactional analysis

Edge processing/analytics is expected to proliferate rapidly in the IoT space due to reduced latency, faster time to insight, reduced storage and bandwidth costs, and (potentially) improved security.


Who Ya Gonna Call?


For edge-based IoT data storage, processing and analysis, more than half (61%) of the 451 Alliance companies rely on cloud providers that offer edge-as-a-service functionality. Examples include AWS Greengrass, Microsoft’s Azure IoT Edge, Google’s Cloud IoT Edge and IBM’s Watson Edge Analytics services.

However, some companies (19%) rely on telecom service providers such as Verizon and AT&T, which offer multi-access edge computing services. An equal percentage of companies rely on their existing on-premises server/gateway providers (e.g., Dell, HPE, Cisco).

A number of startups are also tackling edge analytics. Examples include Nebbiolo Technologies, Crosser Technologies and FogHorn.

Overall, AWS and Microsoft are the most widely used IoT vendors among participants in the 451 Alliance community.

Rounding out the top 10 IoT suppliers are (in descending order):

  • Cisco
  • Google
  • IBM
  • Dell Technologies
  • Oracle
  • SAP
  • HPE
  • Siemens

‘Other’ vendors mentioned include companies that tend to specialize in specific vertical markets (e.g., Verizon, Schneider, Ericsson and Nokia).


Hurdles to Adoption of IoT


All organizations face hurdles on the path to full IoT deployment, particularly challenges associated with capturing and analyzing huge amounts of data.

As usual in the IT arena, cost is a key impediment, particularly in the case of storage resources and network bandwidth. Both of those factors are related to the #1 barrier: IoT simply generates too much data to capture, store and analyze efficiently, which in turn leads to complexities in data cleansing and management.

Barriers to Capturing/Analyzing IoT Data

Storage Woes


Due to the sheer volume of data associated with large-scale IoT deployments, storage stands out as a particularly daunting challenge. On average, IoT data consumes 45% of enterprises’ total storage capacity. At the extreme end of the spectrum, 16% of the 451 Alliance companies report that IoT data takes up more than 75% of their total storage capacity!

About half of the companies expect IoT-related storage capacity requirements to surge by more than 25% over the next 12 months – with an average increase of 35%, while 13% of the companies expect capacity surges of more than 75%!

The storage problem is exacerbated by the fact that the majority (67%) of organizations archive their IoT data after initial capture and analysis. Most (77%) plan to archive the data for two to 10 years, while 15% plan to retain it for more than 10 years.

To mitigate storage costs, companies are increasingly turning to SaaS/IaaS/PaaS cloud providers for their storage requirements – yet another reason why public cloud providers have climbed to commanding positions in the IoT vendor hierarchy.

(For more information on IoT platforms, see IoT Trends: To the Cloud and On the Edge.)