How does software-defined infrastructure make NFVi better?
Telecom operators are investing in network functions virtualization (NFV) to improve speed, efficiency, and agility. NFV is a key technology required to realize the emerging 5G and IoT business potential. At the same time, software-defined infrastructure (SDI) is gaining traction in the market. What is the relationship between SDI and NFV infrastructure (NFVi), and how does NFVi benefit from SDI?
What is NFVi?
NFVi is the infrastructure on which virtual telecom applications (virtual network functions) such as virtual EPC, virtual IMS, and network-near IT applications are deployed. Some examples of network-near applications are tools for performance and configuration management and business support systems (BSS). NFVi includes the compute, storage, and network resources that are always needed in a cloud. In addition, it contains the virtual infrastructure manager (VIM), which might be based on OpenStack, the hardware infrastructure management software, software-defined networking (SDN), and resource orchestration.
SDI and NFVi
So where does SDI fit in the equation? Before we answer that question, here is a brief recap. SDI provides the capability to control and configure compute, storage, and networking hardware from a central management platform into virtual performance-optimized datacenters (vPODs) using an advanced hardware management system.
A vPOD is a dynamically created collection of interconnected hardware built from a pool of hardware resources and can be used for any workload just as with a dedicated POD. The bare metal applications or virtualization layer and applications are then provisioned inside a vPOD. vPODs can be created with different combinations of compute, storage, and networking resources depending on the infrastructure requirements of the applications. An important enabler for SDI is Intel® Rack Scale Design.
See the illustration below showing the NFVi architecture.
How does SDI benefit NFVi?
Flexibility, utilization, and automation capabilities are some of the most important contributors from SDI to make NFVi work better—but how?
Flexibility and resource utilization: vPODs make NFVi more flexible because they can be adapted for specific needs, such as when business conditions change. If a virtualization layer needs more compute capacity, more can easily be added on the fly from the pool of available compute systems. Conversely, if the capacity need decreases, the management system can release resources and assign them to another vPOD instead. This approach helps optimize overall hardware utilization in the datacenter. Being able to segment the hardware pool into vPODs also means that the shared datacenter infrastructure and hardware pool can be used by multiple VIMs without them interfering with one another. Most telecom operators need support for multiple versions of the virtualization layer for various reasons, meaning that they need solutions that support a multi-VIM environment.
Automation: SDI can be used to automate actions, both repetitive operations tasks, as well as actions based on the monitoring of metrics, such as CPU load and memory usage, to minimize the risk of unexpected failures. Over time, the system gathers a wealth of historical data that can help to refine thresholds and policies automatically. The data can be used to identify components that are underused and make them available for other tasks. SDI can also discover components that are likely to fail due to overuse and schedule preventive maintenance. These automated capabilities make the infrastructure more reliable and reduce the risk of downtime of the system or its parts.
And finally, another great example of automation in SDI is the creation, update, and release of vPODs. But I guess you’ve figured that out already, right?
Sign up for this NFV webinar with Jason Hoffman, Head of Technology at Business Area Digital Services, Ericsson.