Industrial cloud moving from public to hybrid systems
A new survey by ABI Research indicates that manufacturers will move IIoT systems from public cloud to hybrid systems as more storage and processing can be done at the device level.
The industrial cloud runs largely in the public domain currently, but that may be about to change.
Over the next few years, manufacturers will move industrial cloud deployments from the public cloud to hybrid cloud systems, according to a new report from ABI Research, an Oyster Bay, N.Y., research firm that specializes in industrial technologies. Public cloud accounts for almost half of the industrial IoT market share in 2018 (49%), while hybrid cloud systems have just 20%. But by 2023 this script will flip, according to the report, with hybrid cloud systems making up 52% of the IIoT market and public cloud just 25%.
The U.S.-based report surveyed vice presidents and other high-level decision-makers from manufacturing firms of various types and sizes, according to Ryan Martin, ABI Research principal analyst. The main focus of the report was IoT industrial cloud and it surveyed the manufacturers and their predisposition to technology adoption.
According to the report, the industrial cloud encompasses the entirety of the manufacturing process and unifies the digital supply chain. This unification can lead to a number of benefits. Companies can streamline internal and external operations through digital business, product, manufacturing, asset and logistics processes; use data and the insights generated to enable new services; and improve control over environmental, health and safety issues.
Changing needs will drive move to hybrid systems
Historically, most data and applications in the IoT resided on premises, often in proprietary systems, but as IoT exploded the public cloud became more prevalent, according to Martin.
The cloud, whether public or private, made sense because it offers a centralized location for storing large amounts of data and computing power at a reasonable cost, but organizational needs are changing, Martin said. Manufacturers are finding that a hybrid approach makes sense because it's better to perform analytics on the device or activity that's generating the data, such as equipment at a remote site, than to perform analytics in the cloud.
You don't want to be shipping data to and from the cloud every time you need to perform a query or a search because you're paying for that processing power, as well as the bandwidth.
Ryan Martinprincipal analyst, ABI Research
"There's a desire to keep certain system information on site, and it makes a lot of business sense to do that, because you don't want to be shipping data to and from the cloud every time you need to perform a query or a search because you're paying for that processing power, as well as the bandwidth," Martin said. "Instead it's better to ship the code to the data for processing then shoot the results back to the edge. The heavy lifting for the analytics, primarily for machine learning types of applications, would happen in the cloud, and then the inferences or insights would be sent to a more localized server or gateway."
Providers like AWS and Microsoft Azure will likely carry the bulk of the cloud load, according to Martin, but several vendors will be prominent in providing services for the industrial cloud.
"There will be participation from companies like SAP, as well as more traditional industrial organizations like ABB, Siemens, and so forth," Martin said. "Then we have companies like PTC, which has recently partnered with Rockwell Automation, doing aggregation and integration, and activation to the ThingWorx platform."
The hybrid cloud market for IIOT will double by 2023.
Transformation not disruption
However, companies face challenges as they move to implement the new technologies and systems that comprise the hybrid industrial cloud. The most prominent challenge is to implement the changes without interrupting current operations, Martin said.
"It will be a challenge to bring all these components like AI, machine learning and robotics together, because their lifecycles operate on different cadences and have different stakeholders in different parts of the value chain," Martin said. "Also they're producing heterogeneous data, so there needs to be normalization of mass proportion, not just for the data, but for the application providers, partners and supplier networks to make this all work."
The overall strategy should be about incremental change that focuses on transformation over disruption, he explained.
"This is analogous to change management in business, but the parallel for IIoT providers is that these markets in manufacturing favor those suppliers whose hardware, software and services can be acquired incrementally with minimal disruption to existing operations," he said. "We refer to this as minimal viable change. The goal should be business transformation; it's not disruption."
