Telecom IQ
Showing results for 
Search instead for 
Did you mean: 

The Nuts and Bolts of physical infrastructure management in NFV


Author: Jonas Arndt, Chief Technologist - Converged Infrastructure


A Straight and Simple Path


There's another side to Network Functions Virtualization (NFV): a physical side. Physical infrastructure management is an essential, yet often overlooked part of virtualized network function (VNF) deployment and operation. After all, physical hardware supports nearly every virtualized resource.


To save both time and resources, physical infrastructure NutsAndBolts.jpgmanagement should always be a completely automatic, transparent and painless activity. By understanding the basics of physical infrastructure management and using the appropriate tools and methods, all of these goals can be achieved.


Easy and Automatic Physical Provisioning


The process of deploying a virtual machine (VM) for a new VNF shouldn't be hampered by the need to manually provision an operating system onto a blade or to enable specific firmware and BIOS settings. As long as there are blades available, these activities should happen automatically and seamlessly. The orchestration function, for instance, should instantly call the Virtualized Infrastructure Management (VIM) layer. The VIM layer must be tightly integrated with the Physical Infrastructure Manager (PIM).


The utilization of available hardware also needs to be an integrated process. A new switch should be provisioned with pre-defined uplinks. VLANs and networks should configure automatically as soon as the switch is introduced as a new physical resource. The same holds true for storage and compute resources.


Interrelated Health and Metrics


Physical infrastructure health and metrics are closely entwined. Along with pushing down workflows and automating hardware provisioning, data on the current state of hardware needs to be pushed up the stack.


The physical infrastructure supplies a hardware base—including servers, storage and switches—enabling the VIM layer to elevate the virtualized platform to the level where VNFs are deployed. The PIM, meanwhile, needs to monitor the status of key physical resources, such as CPU and memory utilization, packet rate, port loading, environmental conditions and other critical factors. The PIM should also issue immediate alerts whenever a resource fails or crosses a certain predefined threshold.


A robust NFV stack needs to have the VIM and the PIM tightly integrated so alerts and metrics are quickly elevated up the stack. Orchestration and VIM functionality both require fast access to this essential information.


Painless Infrastructure Management and Integration


A PIM like HP OneView provides a software-defined approach to physical infrastructure management, including VIM integration. By abstracting hardware management to templates and groups, the PIM enables experts to pre-configure components with ease on a template level. Operators can reuse this expertise over and over simply by deploying the appropriate templates.


Server profiles


A server profile defines various factors, including connectivity, BIOS settings, boot order, firmware level and unique IDs. Users can directly assign a server profile to a specific enclosure bay. Whenever the operator inserts a new blade into the bay, its provisioning is automatically handled by the PIM, and the blade immediately has the correct firmware pushed to it. The BIOS settings and boot order set, as well as full connectivity to all predefined networks, are also automatically loaded onto the blade. Having BIOS settings properly configured is key to deploying VNFs with low latency requirements. A PIM like HP OneView also offers template-driven configurations for interconnects and enclosures. Simply push in a new switch and it is automatically provisioned with the correct firmware level as well as the exact configuration for uplinks and networks.


Resource managers


HP OneView offers different resource managers for handling specific hardware resources. Each resource has a state machine. Any changes to the resources state are pushed to the State Change Message Bus (SCMB). OneView also comes with an extensive set of REST based APIs—even internal resource managers communicate via REST calls. The API set also includes subscription capabilities to the SCMB as well as a pathway to hardware performance metrics.


APIs for VIM-PIM integration


HP OneView uses APIs to enable tight VIM-PIM integration. Any alerts raised by HP OneView are elevated up the stack and injected into HP Helion services, available for northbound consumption. Provisioning new hardware is also easy, thanks to the automatically created server profiles and the automatic provisioning of compute environments. There are plans to continue these integration efforts in the near future, and to include complete access to all metrics in Helion services northbound APIs.


NFV: Ready for Adoption


NFV marks a core structural change in the telecommunication infrastructure marketplace. NFV will bring cost efficiencies, time-to-market improvements, and innovation to telecommunications infrastructure and applications. Physical infrastructure management, when accomplished seamlessly, efficiently and automatically, plays an important role in making NFV a powerful, flexible and cost-efficient technology that's ready for immediate adoption.


To learn how HP can help with your NFV transformation, explore our family of NFV solutions for communications service providers.

0 Kudos
About the Author


Read for dates
HPE Webinars - 2019
Find out about this year's live broadcasts and on-demand webinars.
Read more
Read for dates
HPE at 2019 Technology Events
Learn about the technology events where Hewlett Packard Enterprise will have a presence in 2019.
Read more
View all