BOSTON -- To support a host of new data and applications triggered by the internet of things, more organizations will push compute power out to the network edge. And while some IT pros see this decentralized compute model as a threat to the public cloud market, others just say it'll reshape it.
"There are people out there who say the edge is the death of cloud," said Richard Villars, vice president of data center and cloud at analyst firm IDC. "I kind of say the opposite. I would say that this only works if we figure out how to bring cloud to the edge."
Both the internet of things (IoT) and cloud computing -- particularly the emergence of what IDC calls "Cloud 2.0" -- were big themes at the IDC Directions 2017 event here this week.
By 2019, organizations will process 43% of their IoT workloads on hardware that sits at the edge of a corporate network, rather than in a single or centralized location like a data center, the firm predicted. The lure of edge computing is that organizations can process data as close as possible to its origin, which reduces latency and increases response time.
But even if that means IoT adopters will send fewer workloads to massive data centers -- including those operated by public cloud providers like Amazon Web Services (AWS) -- it doesn't signal doomsday for the overall cloud market, Villars said.
Instead, cloud computing will take on a more distributed feel to bring agile resource delivery and other benefits to those edge locations. The combination of edge computing and cloud is sometimes referred to as fog computing.
"You are going to see cloud not just be in giant data centers, but everywhere," Villars said.
IDC calls these distributed clouds for IoT environments "local clouds." Rather than act as a massive cloud data center that serves thousands of users, these local clouds support a cloud application and all its necessary resources in a specific location to serve a specific set of local users or local machines, he said.
AWS Snowball Edge is an example of a device that fits this model. Users can deploy Lambda code onto the Snowball Edge appliance to capture and process data locally, according to AWS. They can also ship it back to an AWS data center, where the cloud provider adds the data to a central repository for further analysis.
"[Local clouds] are about extending the cloud out, [and] not bringing it back into the enterprise data center," Villars said.
IoT not the only driver for distributed cloud models
Enterprises might choose to pursue a more distributed cloud deployment for reasons beyond IoT. Multicloud -- the use of multiple cloud providers -- is another example of how cloud workloads have become decentralized from a single data center.
Richard Villarsanalyst at IDC
And that trend is only set to continue. By 2018, IDC predicts 85% of enterprises will employ a multicloud strategy. In addition to benefits related to redundancy and disaster recovery, the use of multiple cloud providers allows enterprises to tap into a wider range of infrastructure and platform services than they might get from a single vendor.
"There's going to be a much more distributed set of resource partners that [enterprises] want to take advantage of," Villars said.
Sabre, a provider of global distribution systems to the travel industry based in Southlake, Texas, is an example of a multicloud company. The firm runs its internal SAP systems on Virtustream -- a cloud provider that specializes in SAP workload management, monitoring and support -- while it uses AWS to host its customer-facing apps, as well as for application development and quality assurance.
"I'm trying to keep ... our commercial software that is externally facing toward our customers separate from our internal systems," said Steve Strout, senior vice president of strategy and operations at Sabre.
Another trend that supports the distributed cloud model is industry clouds -- cloud platforms built for a specific vertical market, often by an organization within that particular market. A hospital, for example, might build a cloud platform that offers services that are compliant with the Health Insurance Portability and Accountability Act for other organizations in the healthcare space.
General Electric's Predix cloud, which offers platform as a service tailored specifically toward IoT, is a common example of this model. The U.S. Department of Agriculture's National Information Technology Center also rolled out a government-focused cloud platform that Scott Lundstrom, group vice president and general manager of software, health and government at IDC, described as essentially "FedRAMP as a service."
By 2018, IDC expects more than 450 industry clouds to crop up worldwide -- triple the number running now.
Kristin Knapp is senior site editor for SearchCloudComputing. Contact her at firstname.lastname@example.org or follow @kknapp86 on Twitter.
How cloud adds value to edge computing and IoT
Build a load-balancing strategy for multicloud
Mitigate risk in an IoT deployment