OCP at the edge is predicted to be a larger market than hyperscale. That's quite a statement considering hyperscale capacity is expected to double in the next few years.

To support the growth and maintain the increasingly critical edge, we must build resiliency through a collaborative ecosystem. This was the topic of my presentation at the OCPSummit19.

Reduced latency and the need to process massive volumes of data are driving the edge. Plus, new 5G technology will depend on it. Applications and closed services are moving to the regional edge — some even closer to the user at the local edge.

Call them cloud stacks, cloudettes or outposts, complex architectures including these local edge data centers will be required. Telco networks will build similar architectures with a multi-access edge computing (MEC) model allowing delivery through fiber, radio network and WiFi.

And though users won't know where an application is hosted, they'll surely feel the effects when processing is slow or when the edge goes down. That's why we need to rethink resiliency at the edge. Availability of smaller data centers now has a bigger implication.

Overcome challenges at the edge with these three musts for resiliency

The many challenges of edge come from its many locations and a lack of resources. To effectively and efficiently manage edge and ensure resiliency, you must have the following:

1. An integrated ecosystem that involves physical infrastructure vendors, system integrators, IT equipment manufacturers and managed service providers. Open access will be critical to bring the data to all the parts of the ecosystems. The industry also needs to work towards fully integrated modular micro data centers for the edge that include all IT hardware, security, environmental monitoring, rack access and physical infrastructure.

2. Management tools that are cloud based and can overcome the lack of staff in each edge location. Such a system should keep the sites secure and give you advanced warning before things go wrong. Each system should be managed as a complete micro data center not individual components. Conventional tools can't do this. Having an app, website, anywhere access, pay-as-you-go option, maintenance free, up-to-date cyber security and auto updates are features of a modern tool.

3. Analytics and AI including a solid foundation made up of cloud architecture, data lake and subject matter experts to properly aggregate data securely, create specific AI use cases, identify critical variables, normalize data sets and provide enough compute power to analyze. AI helps create benchmarks, determine patterns and generate alerts to avoid downtime more efficiently than humans ever could.