With the onset of the digital age, the need for agility has become paramount. The network connectivity offered by telecoms is no longer a premium service, and instead has become an on-demand service with instantaneous setup and tear down. The premium service is the set of services offered on top of the network as an application. With the need for flexibility paramount, all of that purpose-built equipment, which was a strength in the past, has become the biggest hurdle to competing in the digital age, contributing to decreasing revenues and increasing cost.
So change was a necessity, and thus was born Network Functions Virtualization (NFV). NFV makes it possible to do most of what the telcos were doing using specialized hardware, but with disaggregated network functions made up of software that could be adapted for new situations on COTS hardware.
All of this requires fast evolving infrastructure resource management, of course, and for that reason, NFV has become virtually inseparable from the OpenStack cloud platform. Mirantis, along with partners and competitors, has been working steadily with early adopters to make sure that OpenStack has what it needs to be well suited for NFV deployments. In particular, for the past several years, Mirantis has worked within the telco community to support adoption of OpenStack as the primary Network Functions Virtualization Infrastructure (NFVi) and Virtual Infrastructure Management (VIM).
These early adopters of NFV and SDN for telco network transformation have spoken about the initial successes of this approach, as well as the areas that need to be addressed in order for NFV to reach the next level — in order words, massive adoption across the globe.
Some areas our customers and partners identified as crucial include:
- CI/CD for the infrastructure layer: One of the biggest hurdles to NFV is that NFV and its ecosystem are continuously evolving, so operators need a proven path to seamlessly absorb new innovations into every component of NFV, including the infrastructure layer. To solve that problem, we need to build Infrastructure as Code to enable infrastructure lifecycle management (LCM).
- Future proof the infrastructure layer: It’s not enough to be able to manage the infrastructure; we need to make sure that we’re avoiding the need for forklift upgrades when major changes come along, such as the move to support container based, cloud native VNFs.
- End-to-End automation including VNF onboarding and monitoring: This is a key requirement for business agility which is critical for lowering time to market and revenue acceleration. It enables optimal resource utilization and prevents stranded/stolen assets.
- Strong open source communities: No single organization can afford to innovate at speeds essential for the transformation of extremely complex telco network infrastructures, so our customers recognize the need for strong community support for the components of the NFV architecture, especially the management and orchestration (MANO) and virtualization layers of NFVI within the ETSI NFV reference architecture.
Our customers include service providers all over the world, so these problems have been top of mind for us for some time, and we’ve been working to solve them. For example:
- Mirantis Cloud Platform (MCP) includes DriveTrain, which provides a platform for managing virtualized networks using infrastructure as code
- In addition to DriveTrain, MCP includes the ability to easily add Kubernetes and containers, making it an ideal future-proof platform for telcos. What’s more, DriveTrain makes it possible to add the “next big thing” in a manageable way.
- Mirantis is actively contributing to both ONAP and OPNFV, and currently working on solutions for VNF onboarding and monitoring.
- Mirantis is an open source company, and as such, all the components are built on open source tools, so service providers can lean on a global pool of resources for innovation in the infrastructure area.
In fact, the latest version of MCP focuses on the specific needs of NFV workloads, including their operationalization, or orchestration and automation within the context of a telco network. For example, MCP includes:
- Capacity management of SR-IOV NICs through QoS controls. Bandwidth capping on a per virtual function level permits fine-grained traffic shaping and prevents noisy-neighbor syndrome.
- Better reliability, higher bandwidth, and improved load balancing with OVS-DPDK support on bonded NICs. This also enables operators take advantage of existing assets. For example, you can utilize 10G NICs, when available, instead of investing in 40G NICs.
- Improved performance for DPDK by pinning individual queues to cores with NUMA affinity
- The ability to run telco VNFs that require simultaneous connectivity to multiple networks through support for VLAN aware VMs.
One thing that we know for certain is that telcos and service providers can’t afford to ignore NFV. As addicted to their phones as many people are now, the “unlimited” network capability that is expected with the upcoming 5G standard has the potential to make connectivity seem like it is the fourth essential ingredient for human survival (after water, air and shelter). And as complex as 5G will be, NFV is critical for it to become a reality. 5G requires a dynamic hierarchical architecture; between that and requirements for network slicing, and cloud-based radio access networks (C-RAN), virtualization of the networking infrastructure is essential.
Accordingly, Mirantis has a rich roadmap that focuses heavily on NFV for 5G enablement over the next 6 to 18 months, and we can’t wait to share it with you.