One of the defining characteristics of edge applications is the need for low latency to absorb and analyze data from connected devices deployed in locations such as retail stores, manufacturing facilities, distribution centers, and municipal infrastructure.
Until recently, most chatter about “the edge” has been vague, often conflating the extension of cloud service delivery to remote locations with more precise scenarios such as Multi-Access Edge Computing (MEC). MEC combines low-latency 5G access networks with cloud services.
That scenario—bringing cloud services close to users and applications with sub-millisecond latency—is necessary for many Industry 4.0 and near-real-time scenarios. As I detailed earlier this year on the Packet Pushers’ Ignition site, it was also the motivation for Vapor IO, VMware, and others to form the Open Grid Alliance (OGA). These companies are addressing the growing need for low-latency distributed infrastructure through a combination of
- Micro-data centers strategically positioned with wireless basestations,
- A new generation of edge-optimized network and system management software
- Tight integration with cloud service providers
As Vapor IO’s founder and CEO Cole Crawford wrote in announcing the OGA (emphasis added),
“The edge is also at risk of being balkanized as multiple projects implement their own solutions on single-tenant infrastructure with isolated resources. We, as an industry, cannot deliver the applications of the future unless they truly can run anywhere, on demand. We need a frictionless edge with resources that can be called up as needed to solve specific problems. In the Open Grid world, when applications demand resources—in a particular place, at a particular time, with particular SLAs—the grid will deliver that, on demand.”
Source: Google Trends shows growing interest in edge computing in contrast to higher, but static curiosity about Industry 4.0.
INZONE: The Open Grid Embodied
Only four months after the OGA announcement, Vapor IO has built the first implementation via its new INZONE program. The program includes a set of prefabbed modular data centers positioned at select 5G base stations to provide coverage across a metropolitan area. Essentially INZONE combines Vapor IO’s Kinetic Grid (detailed in my Ignition column) with a set of partner-provided infrastructure and application services to provide what Vapor IO describes as “near-premises” locations that combine the low-latency performance of on-premises systems with the reliability, security, scalability ,and service portfolio of a cloud region. As Crawford puts it (emphasis added),
“It all comes down to delivering infrastructure within microseconds of the target facilities, then enabling that infrastructure with last-mile networks combined with the intelligence of software APIs and real-time telemetry. We call this unique combination of technologies INZONE because it enables Industry 4.0 application within a physical geography without requiring on-premises data centers.”
Source: Vapor IO
Las Vegas is home to the first INZONE deployment, which starts with three Kinetic Grid hubs located near key commercial zones. Each site is positioned to provide 100µs latency via fiber or wireless within a 10km radius (approximately 120 square miles).
As illustrated by the following map, locations with overlapping coverage have redundant access to two or more INZONE sites, allowing the Kinetic Grid software to automatically migrate and restart workloads in another site yet maintain the sub-0.1ms SLA. Phase II of the Las Vegas project will add three more sites, significantly expanding the service area for both single-site and redundant coverage (note: I added a red outline to highlight the area with redundancy).
Source: Vapor IO; annotation by the author.
Infrastructure Is Useless Without Services
As my earlier Ignition article noted, “Kinetic Grid’s competitive differentiation [lies in] cloud and network neutrality in cloud-quality modular data centers that can be rapidly deployed in any location with available Zayo fiber.” While Kinetic Grid provides the edge infrastructure, the point of INZONE is pairing its network and facilities with equipment and cloud services for latency-sensitive applications. So far, there are eight companies providing INZONE services, notably AWS, Turbine, and VMware.
- AWS is hosting Outposts, 1U and 2U (INZONE doesn’t yet support the full-rack systems) appliances that can host a subset of AWS services including EC2 instances, ALB load balancing, EBS and S3 storage, ECS and EKS container clusters, RDS databases and EMR data analytics.
- VMware offers its Telco Cloud Platform which provides 5G, RAN, SD-WAN and other network services.
- Terbine provides IoT data collection, exchange, monetization and analytics to support applications in AR, AV guidance, traffic management and smart municipal infrastructure.
Source: Vapor IO
A compelling enterprise application for INZONE is for provisioning private 5G (via CBRS spectrum) using software-defined vRAN to run the DU and vCU on Outposts or bare metal using Telco Cloud or similar RAN software. Because INZONE can reach any 5G tower in its coverage radius within 100µs, it provides enterprises with 5G performance without requiring them to install RF or RAN equipment. Indeed, the city of Las Vegas will be using INZONE to expand coverage for an existing program using private 5G to provide Internet access for those lacking access to cost-effective broadband service.
INZONE fulfills Vapor IO’s potential to become a cloud- and carrier-neutral hosting provider for edge infrastructure and services. It already has the national and metropolitan connectivity in 36 markets and, according to Crawford, has done the regulatory and site-selection legwork to allow rapid expansion of INZONE deployments nationwide. I look forward to seeing how rapidly it can attract the enterprise customers and service partners necessary to expand INZONE availability to other markets.
