The edge is everywhere: in the media and at trade shows, in analyst reports and business plans. Everyone sees it coming, and no one can afford to ignore it.
But at the same time, edge computing is still emerging in terms of live production deployments that move beyond central and regional data centers and out to telco central offices or 4G/5G base stations.
Today we announced several key innovations that are designed to change that and help push the edge from vision to reality.
Before I dive into the solution, let’s define the problem.
The Demanding Edge Environment
There are good reasons why edge deployments in central offices and at 4G/5G base station locations are slow to take off. For one, edge computing requires a fundamental rethinking of how to build and automate highly distributed cloud networks. It requires purpose-built hardware and software that is easy to deploy and manage while still meeting the more stringent constraints of these mini data centers. To date, most vendors have tried unsuccessfully to retrofit technologies and products designed for large-scale centralized data centers and found that they just don’t fit the significant space, power, environmental and cost constraints of edge locations such as telco central offices and 4G/5G base stations, let alone even smaller sites in street cabinets or harsh rural environments. And while industry organizations have published blueprints and reference architectures for edge networking, most of these are still at the stage of science experiments in the lab or, at best, in “proof of concept” field trials.
What are the requirements for networks in remote, lights-out edge data centers?
What we need:
- Comprehensive automation and integration with orchestration platforms
- Low-latency distributed architectures that interoperate with existing network infrastructure
- Deep network slicing and segmentation for multi-tenant security
- Comprehensive visibility across edge locations with strong analytics
- Carrier-grade, resilient hardware/software solutions, often including NEBS compliance
What we cannot afford:
- Lots of rack space
- High power consumption
- High capital cost per site
- Manual processes and high OPEX
Traditional data center network automation solutions have been designed for central data centers and based on a centralized controller model. Thus, getting all the required automation, segmentation, visibility and analytics means adding more and more hardware and software components at every edge site – SDN underlay controllers, virtualization overlay controllers, virtualization overlay licenses for every server CPU, network taps/probes, analytics servers, etc. – and that drives space, power and cost. And because these systems were not designed for distributed environments they require “controllers of controllers” in order to handle more than one site, adding even more cost and complexity. That’s why those traditional data center networking solutions just don’t fit at the edge.
Edge Network Automation. Simplified.
How are we addressing this problem? Today we announced several new solution components that build on our distributed cloud solution, announced last year. Let’s look at a few of the key new pieces and how they contribute to create a complete edge network automation solution:
- First Central Office Edge Switch: Celestica Edgestone is the industry’s first NEBS-compliant top-of-rack switch specifically designed for central office (CO) edge deployments, and Pluribus is the first to deliver network operating system software to support it. It has been clear for years that edge computing will come to central offices, but Celestica and Pluribus are now delivering a carrier-grade, production-ready network solution to make it real.
- DC Gateway Router Integration: Edgestone is not only CO-friendly, it’s powerful, with a Broadcom Trident 3 NPU, an Intel Broadwell 12 core CPU with plenty of memory and a high-performance control plane interface between them leveraging dual 10G interfaces connecting them. Pluribus takes advantage of all that power to run not only the Netvisor® ONE OS but also to integrate data center gateway router functionality that previously needed to run on a separate hardware platform. When we launched the data center gateway router last summer it was the world’s first multi-tenant white box data center gateway router. Integrating data center gateway functionality into the top-of-rack switch delivers another “world first” and, more importantly, it further reduces the hardware space, power and cost required for network functionality in edge locations.
- Red Hat OpenStack Integration: Automation at the edge is not optional. It’s fundamental. And we are taking edge network automation to the next level.
At the core of the Pluribus distributed cloud solution is the Adaptive Cloud Fabric, which is critical to creating edge network fabrics that are fully automated and radically simplified. This unique controllerless software-defined networking (SDN) solution does not require the typical three controllers at every edge site because the software is distributed and runs as an application on the leaf switches that are being deployed for connectivity anyway. Thus, the Adaptive Cloud Fabric incorporates network automation, network virtualization, network segmentation, visibility and analytics in a much more comprehensive, cost-effective and easy-to-use solution compared to traditional solutions, and it also interoperates with existing network infrastructure and is ideally aligned to highly distributed edge networks.
But of course, networks do not exist in isolation, and that’s why we have embraced integration with NetOps automation tools such as Red Hat Ansible and other orchestration solutions like VMware vCenter. Now, to address the requirements of leading telecom network operators and cloud service providers, we have added integration of Netvisor ONE with Red Hat OpenStack R13 and expanded our Red Hat Ansible playbook library.
Red Hat OpenStack is deployed in a large number of telecom providers today, so it’s the right choice right now for real edge network deployments. As we look toward telecom providers embracing containers in their future production networks, we are extending our integration to support Red Hat OpenShift and Kubernetes. Additionally, Red Hat OpenStack R13 is completely containerized itself and is integrated with OpenShift, paving the way for a smooth transition to a world where both virtual machines with OpenStack and containers with OpenShift/Kubernetes will coexist for a number of years.
When you put all these pieces together, it’s clear that we are delivering a comprehensive, automated, and simplified solution for edge networking that no one else can match.
Looking for your edge? It just got real. Contact us to learn more.