Micro Focus is now part of OpenText. Learn more >

You are here

You are here

OpenStack: Now for building clouds at the edge

public://pictures/jasonbloomberg-tight-300x412.jpg
Jason Bloomberg President, Intellyx
 

When I first wrote about OpenStack in 2015, I described the open-source, private-cloud software effort as "the mother of all learning opportunities." At that time, too many vendors had jumped on the OpenStack bandwagon, leading to a proliferation of poorly integrated projects and a dearth of enterprise deployments.

I recently attended the OpenStack Foundation’s freshly rechristened Open Infrastructure Summit (OIS), known as the OpenStack Summit in days past. In spite of the optimistic name, this conference was notably smaller than previous ones—and the number of vendors driving the effort had similarly slimmed down.

The conference's smaller size, however, belied the level of support for OpenStack, or in any case, for some of its numerous projects.

Across industries, enterprises reported extensive production deployments of core OpenStack projects, including Swift and Cinder for storage, Neutron for networking, and Nova and Ironic for compute.

Many of the other projects, in contrast, are now languishing out of the limelight, or playing relatively minor support roles to the storage, networking, and compute capabilities that make up the essential elements of any private cloud.

The success of core OpenStack projects in the enterprise, however, wasn't the most exciting takeaway from OIS. The real news is what the telecommunications industry is doing with OpenStack—especially at the edge.

Here's what your team needs to know about today's OpenStack.

Telecoms build the edge

Telecoms see "the edge" a bit differently from the traditional IT perspective. According to the telecoms, there are actually three different edges. First, there's the "cloud edge," where the content delivery networks (CDNs) play. CDNs locate servers close to end users to distribute traffic and reduce latency. A CDN's edge server is thus a web server that fetches data from a master server and caches it locally at a data center at the edge of the cloud.

[ Partner resource: Subscribe to Intellix's Cortex and Brain Candy Newsletters ]

Edges near and far

In addition to the cloud edge, there are the "near edge" and the "far edge," the combination of which is the "broad edge," according to Glenn Seiler, vice president of product strategy at Wind River.

The near edge consists of servers located at cell towers or local telecoms points of presence. Such locations are typically well-suited to host small data centers with all the essential trimmings—power, cooling, racks of equipment, and sometimes a bare-bones staff.

The near edge may also represent a facility that hosts Internet of Things (IoT) gateways, perhaps at a factory, in a city's traffic switching office, or in the security command center of a large building (or campus) such as a stadium, airport, or office building.

The far edge, then, consists of the devices themselves, either handheld smartphones or IoT sensors and actuators.

Today, the telecoms are putting OpenStack (as well as Kubernetes) at near-edge facilities. Simply having clouds in traditional, centralized data centers isn't sufficient. The reason? 5G.

5G and OpenStack: What you need to know

The 5G rollout is just getting started and is likely to take a decade when all is said and done. But for the telecoms, building out the 5G infrastructure is already job No. 1.

5G, however, isn’t just one type of service. In reality, it consists of a family of services at different wavelengths that work at different distances. 5G will also support a wide range of devices at the far edge—not just phones, but IoT devices in far greater number and variety than we're familiar with today.

And everything, regardless of which edge or which service, will go much faster than ever before. Just one problem: the pesky speed of light. The farther a message travels, the higher the latency—and latency slows everything down. The solution: Bring clouds to the near edge.

If your driverless car has to communicate with a centralized cloud, the latency will be too great for a cloud-based app to provide information to the car in time for it to take immediate action. But if the cloud is at a cell tower 100 meters away, say, then the round-trip 5G signal will be plenty fast enough.

OpenStack rises to the challenge

To meet the needs of the telecoms at the edge, as well as enterprises with similar problems, the OpenStack community is bringing several new projects to market. For example, StarlingX is a "complete cloud infrastructure software stack for the edge used by the most demanding applications in industrial IoT, telecom, video delivery and other ultra-low latency use cases," according to the project's website.

Complementing StarlingX is the newly released Airship 1.0, which is a "collection of open source tools for automating cloud provisioning and management," according to its website. Airship works well in traditional private clouds, but also provides simplified provisioning and configuration for the near edge—and it's enterprise-ready and powering 5G today, according to the OpenStack website.

Another near edge must-have: bare-metal provisioning. Instead of installing OpenStack in a virtualized environment, there are many situations where a company might want to install it on hardware that has no pre-existing software on it.

Bare metal in the stack

To meet the bare-metal needs of both telecoms and enterprises, the OpenStack community has been building out several projects—in particular, Ironic, one of the most talked-about projects at OIS.

There are several reasons to install OpenStack on bare metal:

  • For high-performance computing clusters
  • For tasks that require access to specialized hardware
  • For rapid deployment of cloud infrastructure
  • For single-tenant deployment for performance, security, and dependability concerns
  • For particular regulatory requirements

The telecoms have several use cases for bare-metal OpenStack, but supporting Network Functions Virtualization (NFV) is central to the story. With NFV, generic hardware can serve the role of any device in the telco infrastructure via a fully software-defined approach. OpenStack is rapidly becoming the cloud platform of choice for NFV.

NFV, however, isn’t the whole story for the telecoms. "On top of bare metal we have virtual machines, Kubernetes, containers, and much more," said James Penick, architecture director at Verizon Media. "Our production workloads run at all levels of this stack, and my dream is to one day push as much of our workloads into higher functions as much as possible."

[ Also see: The 4 stages of an IoT architecture ]

Advice for practitioners

For developers and operations personnel who are looking to get into OpenStack, the sheer number of projects and other moving parts can be daunting. The good news: Your organization is likely to have a core set of use cases for the technology. Once you understand what problems you're trying to solve, the path becomes far simpler.

For more senior technical staff, it's essential to balance the day-to-day operations of OpenStack today with the ongoing innovation that is laying out a path for the future. OpenStack at the near edge is at the epicenter of such innovation, since it brings together IoT, 5G, and one of the primary enterprise use cases for edge clouds: artificial intelligence inference.

If you're on a technical team at a telecom​​​​, the story today is all about 5G. But never forget, there will be a multi-year transition period when previous-generation technologies must run alongside the new ones.

OpenStack has an exciting role to pay in tomorrow's 5G world, but it's also today's de facto standard for private clouds—and its penetration both in telecoms and enterprises is only going to grow. 

Disclosures: None of the organizations mentioned in this article are currently Intellyx customers. The Open Infrastructure Summit provided Jason Bloomberg with a free pass to the event, a standard industry practice.

Keep learning

Read more articles about: Enterprise ITCloud