Blog co-authored by Delmar Hernandez, Senior Principal Engineer.
Communications Service Providers (CSPs) began their virtualization journey approximately 10 years ago with the initial move of core networks to private clouds. This move to an x86-based, data center typical server environment was a departure from a focused purpose, customized hardware platforms. Instead, rack designs were dimensioned and standardized for multiple Virtualized Network Functions (VNFs) to co-exist in the same cloud, on common hardware, where multiple purpose-built hardware solutions were previously required.
With this transition complete in the core network, attention has shifted to the Radio Access Network (RAN), with the same promise of migration from multi-vendor, purpose build hardware solutions to a common base hardware platform, with a cloud abstracting the details of the underlying hardware. This will maintain a competitive, multi-vendor RAN software ecosystem and allow CSPs to decouple hardware from software to create even more vendor competition and opportunities to select best-in-class solutions.
The O-RAN Alliance has taken up the quest to transform the RAN into an open, intelligent, virtualized and multi-vendor interoperable network. While the O-RAN Alliance does call out Commercial-Off-The-Self (COTS) servers/storage/networking for regional cloud deployments, a very different environment can exist at the edge.
Typical Telecom Edge Data Center (DC) deployments scale up from the site support cabinets and concrete huts, that you’ll typically see at the base of most every cell site in more suburban and rural deployments, to micro and container DCs. These enclosures scale from a few rack units (RU) to several or even hundreds of racks. These sites typically have the same infrastructure components that you find in a large data center. All the components of a traditional data center were scaled down to a smaller, more focused deployment. However, the availability (tiering as defined by the Uptime Institute) of the site is not on par with the larger data centers due to the cost of providing full redundancy for so many distributed sites. These sites are typically designed in a Tier 1 or Tier 2 level, with expected uptimes in the 99.6-99.7% range, or about 1.5 days of outage per year.
Of course, with the increasing densities and redundancies of more data center-like environments comes an increase in infrastructure costs. This is the balance that telecom operators struggle with when planning to deploy an edge compute infrastructure.
Under certain failure conditions, such as HVAC failure or weather variations, telecom equipment can be exposed to temperature extremes for an extended period of time, either until a repair is affected or temperatures moderate. As a critical resource in times of emergency, telecom networks must be designed to handle extremes in environment, and the equipment being deployed to create edge clouds are no different. This necessitates that compute resources deployed into edge cloud conform to the design specifications of similar, existing, purpose-built hardware previously deployed.
This set of design specifications for electronic equipment being deployed into telecom environments is collectively called the Network Equipment Building Systems (NEBS) specifications. The details of NEBS specifications are not widely understood. Still, the results of NEBS compliant hardware designs are well understood in the Telecom Industry for their ruggedness and ability to operate safely in harsher environments that would typically break or dramatically shorten the operational life of an Commercial-Off-The-Shelf (COTS) device.
This is an introduction to our series exploring details of the NEBS specification and design decisions that create platforms that can not only exist in a NEBS specified environment, but often exceed those requirements to provide some buffered tolerance against the extremes of edge deployments. We’ll endeavor to cover the following topics and more:
- NEBS Genesis and Overview
- Physical Protection Testing
- Electromagnetic/Electrical Safety Testing
- NEBS Criteria Levels
- Other Design Considerations for the Edge (multiple parts)
- Dell’s Solutions for Edge Compute (multiple parts)
The goal is to provide a greater understanding of the design and testing efforts required for an equipment vendor to state that a product is NEBS Compliant. In addition, once the topic of NEBS and Edge Cloud deployments is better understood, the reader will be introduced to several of Dell Technologies’ Edge Optimized server offerings and how these servers are designed to meet and often exceed the requirements of NEBS.