NFV Introduction, Telco Use Cases & Architectural Framework

What is NFV?

NFV stands for Network function virtualization. As you know, every operator has lots of equipment and hardware, which each of them are performing specific functions. For example in the scenario of the 4G network, we have eNB, MME, SGW, PGW, HSS, PCRF, DNS servers, and many routers and switches.

When new mobile technologies are developed such as 5G, new hardware changes in the entire network should be done. For example, an operator with 4G if want to provide 5G for customers, many hardware changes and activities are required to swap hardware and redesign a new network.

Currently functions are bind to dedicated hardware, which means for example hardware working as HSS in 4G, cannot work, as HSS of 5G and new installation is required for 5G HSS implementation.

Therefore, NFV is a way to reduce cost and speed up service deployment for network operators by decoupling functions like session management, mobility management, encryption, or authentication from dedicated hardware and moving them to virtual servers.

NFV aims to transform the way that network operators architect networks by evolving standard IT virtualization technology to consolidate many network equipment types onto industry standard high volume servers, switches, and storage.

Instead of installing dedicated hardware, virtual machines are taking the role of network functions. Therefore, if there is a need for a new network function, the operator can simply perform it on VMs as a software.

For example, instead of installing dedicated hardware as HSS for user subscription management, it can be deployed as software on VMs.

Network function virtualization reduces dependency on dedicated hardware and allows higher scalability and customization across the entire network.

What are the advantages of NFV?

NFV reduces the need for dedicated hardware for the deployment of network elements by changing network functions into software than can run on virtual machines on standard servers. Separating network functions from hardware has many benefits for the CSPs and mobile operators. Which include:

• Reduced space needed for network hardware
• Reduce network power consumption
• Reduced network and hardware maintenance costs
• Easier network upgrades
• Longer life cycles for network hardware
• Higher reachability for nodes within the same VM server

NFV standardization

In 2012, seven of the world’s leading telecoms network operators, including AT&T, Deutsche Telekom, Orange, and China Mobile publishes a white paper on network function virtualization. The key objective for this white paper was to outline the benefits, enablers, and challenges for Network Functions Virtualization.

Since 2012, ETSI became responsible for standardization of NFV and community is called ETSI ISG NFV. Until now, ETSI publishes four Release of NFV, each in almost every two years.

NFV Use Cases in Telco

There are many use cases, which are defined for NFV. Here are four of them related to the telecommunication industry directly.

• Virtualization of Mobile Core Network and IMS
• Virtualization of Mobile base station
• Network Slicing
• Virtualization of Internet of Things (IoT)

Virtualization of Mobile Core Network and IMS

What are the important aspects in the virtualization of mobile core networks and IMS?

Mobile networks are populated with a large variety of hardware appliances in their core network. Virtualization has some advantages and challenges. Advantages of the virtualization of mobile core network and IMS include the following:

• Reduced total cost of ownership (TCO)
• Improved network usage efficiency due to the flexible allocation of different network functions on the hardware resource pool
• Higher service availability and sustainability provided to end-users/customers by dynamic network reconfiguration inherent to virtualization technology.
• Scalability: Capacity dedicated to each Network function can be dynamically modified according to actual load on the network, thus increasing scalability.
• Topology reconfiguration: Network topology can be dynamically reconfigured to optimize performances and to support the agile introduction of new services.

What is Virtualization Target in Virtualization of Mobile Core Network and IMS use case?

Virtualized EPC and IMS in network operators are called vEPC and vIMS.In the EPC and IMS, the following network functions need to be virtualized:

• EPC Core and Adjunct Network Functions e.g. MME, S/P-GW, PCRF, etc.
• 3G/EPC Interworking Network Functions e.g. SGSN, GGSN, etc.
• All IMS Network Functions e.g. P/S/I-CSCF, MGCF, AS.

As the mobile core networks already deployed are not based on NFV, the NFV-based virtualized core network should coexist with a non-virtualized mobile core network. This figure shows how a network, which is partially virtualized, coexists with other parts.

There are two most possible scenarios, which operators follow for NFV deployment.

• Partial virtualization of the mobile core network
• Service specific mobile core network virtualization

In this case, only some NFs are Virtualized. They can be EPC control functions (e.g. MME/SGSN), HSS or IMS nodes (e.g. CSCF).

The operator also can deploy a complete Virtualized core network while still having the non-virtualized one. The Virtualized core can be used for specific services and/or devices (e.g. machine-to-machine) or for traffic exceeding the capacity of the non-Virtualized network.

Virtualization of Mobile base station

What are the important aspects of the Virtualization of Mobile base stations?

Mobile network traffic is significantly increasing by the demand generated by the application of mobile devices. To fulfill cellular traffic demand for higher data rates in the future there is a need for a proper amount of RAN nodes. Considering Capex and Opex and continuous installation of new RAN nodes, this part account for most of TCO and power consumption in mobile operators, and virtualization will help to reduce them.

Advantages of the Virtualization of Mobile base station include the following:

• Lower footprint and energy consumption
• easier management and operation
• faster time-to-market
• dynamic resource allocation and traffic load balancing

What is Virtualization Target in this use case?

The virtualization target, in this case, is to create centralized BBU (Baseband processing) Pool for Baseband processing units and functionalities of MAC, RLC, PDCP, and RRC. Then different Radio units Antenna are connected to BBU pool using fiber and communication between nodes can be performed in the same BBU pool more efficiently.

Inside a RAN node, purpose-built hardware might still exist since all the baseband processing functions cannot be efficiently realized on software.

For the coexistence of virtualized and non-virtualized network functions in the mobile base stations, virtualization should support Partial deployment.

In this case, some performance issues may happen due to the virtualization and topological situation. For example in the scenario of eNB consider that there is a need for X2 interface between Virtualized eNodeB in BBU pool and a Non-virtualized eNodeB.

Network Slicing

What is Network Slicing?

The concept of Network slicing consists of running multiple logical networks on a common physical infrastructure. Network slicing is described as a logical network between a set of network devices and some backend applications to deliver services for users or a set of users.

Typical network slice would be for IoT devices to connect to the network and reach backend M2M applications, or another instance for smartphones to connect to the same network but to reach out to VoLTE IMS servers.

The network slicing concept has been described first in 5G white papers.

What are the advantage of Network slicing?

While network slices could be defined on top of physical resources, the benefit of using virtualized resources gives the flexibility inherent to NFV such as

• sharing resources
• allocating resources to a slice dynamically
• scaling automatically
• self-healing
• deploying a slice automatically

The slicing idea is based on the concept that each slice will be created to cover the demand for one or more services. Here are services, which each slice cover:

• providing connectivity between endpoints (terminals or network gateways)
• processing traffic in between end-point where needed
• providing network and Services management capabilities (OSS)
• providing support for the business administration (BSS)

Network slices may have dedicated network functions or multiple network slices that can share the same set of network functions and physical resources.

Several technologies for the realization and provisioning of network slices are already available. Three of them are considered to be of direct relevance for this use case.

• NFV (Network Function Virtualization)
• SDN (Software Defined Network)
• SDR (Software Defined Radio)

These are the key enabler Technologies to enable the slicing concept.

A typical example of the network slicing is setting a network slice between a given application in a connected car and a back end application (i.e. WebRTC or Video streamer). The end-user wants to stream videos he has purchased before his trip. The videos are stored on video streamer in the network, ready to be streamed to the car with proper quality. The car is connected to a cellular network and is driving. It is running an application that is requesting a given frequency band, i.e. LTE, a given bandwidth, end to end quality of service, etc.

Some network slices may have specific requirements such as end to end low latency which need to be taken into account when defining/deploying the network slice, but also when operating the network slice to ensure the proper SLA are met.

Each network slice should have its own network slice manager for orchestration and management of the slice.

What are the virtualization targets in network slicing?

There are two targets in network slicing virtualization, which are virtualizing

• Network functions that compose the network slice
• Network slicing management and orchestration entities (MANO)

Which manages slices for allocation of resource, QoS, etc.

Virtualization may be hypervisor or container-based. A given network slice may support a combination of hypervisor-based resources and container-based resources.

In addition, in-network slicing for the coexistence of virtualized and non-virtualized network functions, Network slices include a mix of virtualized and non-virtualized network functions.

Virtualization of Internet of Things (IoT)

The fourth use case is the virtualization of IoT. IoT applications comprise a large variety of service types, involving a number of players, and encompassing a wide range of requirements. The efficient delivery of services requires the deployment of IoT-related functions over NFV domains, possibly combined with public/private clouds.

IoT virtualization use case is based on an IoT study published in 5G white paper.

What are the advantages of virtualized IoT:

• Efficient utilization of the IoT services
• Agility of automation of new IoT based services

What are virtualization targets in IoT?

All traditional network functions used for providing the Internet of Things (e.g. connectivity, authentication, storage, and gateway functions) are candidates for virtualization.

How about the coexistence of virtualized and non-visualized network functions in IoT?

In fact, IoT could include a mix of virtualized and non-virtualized network functions. For this use case, the coexistence of virtualized and non-virtualized network functions is not anticipated to give rise to specific problems since the functions are generally linked through standard interfaces.

NFV Architectural Framework

The major role of NFV is to decouple network functions from purpose-built hardware and use shared hardware resources for different network functions. To achieve this goal, ETSI as a standard development organization came up with a single high-level framework to align hardware and software providers.

After the implementation of NFV in an operator, it is possible for one mobile operator to implement different network functions provided by different vendors in the same hardware infrastructure. For example, MME of Huawei can work with the HSS of Nokia in the same physical infrastructure.  

Here in the above figure, it shows a high-level NFV framework. As shown in the figure with different colors, there are three main working domains are identified in NFV:

• Virtualized Network Function (VNF), as the software implementation of a network function which is capable of running over the NFVI
• NFV Infrastructure (NFVI), including the diversity of physical resources and how these can be virtualized. NFVI supports the execution of the VNFs.
• NFV Management and Orchestration (NFV-MANO), which covers the orchestration, and lifecycle management of physical and/or software resources that support the infrastructure virtualization, and the lifecycle management of VNFs. NFV Management and Orchestration focus on all virtualization-specific management tasks necessary in the NFV framework.

VNF can host containers or virtual machines with virtualized network functions for example MME in 4G network.

Let’s have an example of virtualization using the NFV high-level framework.

Suppose that there is a traditional 4G network with MME, SGW, PGW, HSS, and PCRF as network nodes, in which each of them is performing a network function.

For NFV deployment, there is a need for high capacity servers, switches, and storages provided by mobile operators. As described, the NFV framework contains three major parts. VNF, NFVI, and NFV-MANO.

NFVI provides virtualization transformation from hardware to VNF. VNF can host virtual machines or containers for each network function and NFV-MANO is used for management and orchestration of physical and virtual resources such as CPU, Memory, and network access.

Each node in the 4G traditional network is a VNF in a virtualized network and is deployed on single or multiple virtual machines or containers.

Now let’s dive deeper into NFV architecture and explain all layers and functional blocks.

The first layer involved with the hardware is called NFVI. The NFVI is the totality of hardware and software components in which VNF is deployed on it. The NFVI can span across several locations that are the place where NFVI-PoPs are operated.

From the VNF point of view, the virtualization layer and hardware resources look like a single entity providing virtualized resources.

Hardware resources

In NFV, physical hardware resources include computing, storage, and network that provides processing, storage, and connectivity to the VNFs through virtualization layer that is for example hypervisor.

This hardware is assumed to be COTS which means commercial off-the-shelf and is not purpose-built hardware and could be provided by HP, Huawei, or other companies.

Computing and storage are commonly pooled. Network resources consist of for example routers, switches, and wired or wireless links. NFV differentiate two types of networks:

• NFVI-PoP network. The network that interconnects the computing and storage resources contained in an NFVI-PoP.
• Transport network. The network that interconnects NFVI-PoPs, NFVI-PoPs to other networks

Virtualization Layer and Virtualized Resources

The virtualization layer decouples VNF from underlying hardware ensuring the independent lifecycle of VNFs.

The use of hypervisor or containerization is two choices for deployment of the virtualization layer. In some cases, VMs may have direct access to hardware resources for example network interface cards for better performance.

For virtualization of a network domain, to ensure virtualized network paths that provide connectivity between VMs and VNF or with other VNFs in other NFVI-PoP, several techniques such as VLAN, VPLS could be used.

To manage the interaction of VNF with virtualized computing, storage and network resources, VIM, which is Virtualized Infrastructure Manager, is used in NFV-MANO.

All assignments of virtualized resources is under the authority of VIM. It also provides operation and reporting. Therefore, VIM is responsible for:

• Allocation of virtualized resources to upper layer VMs or containers
• Inventory management for physical computing, storage and network resources dedicated to NFV infrastructure
• Providing performance analysis
• Collection of events and faults information
• Collection of information for capacity planning, monitoring and optimization

In addition, to note, multiple VIM instances may be deployed in the NFV-MANO.

Virtualized Network Function

A VNF is a virtualized version of the network function in a traditional network. For example, each VNF can host MME, SGW, and PGW functions in the 4G network core.

For partial deployment of NFV in the network, VNFs could work with external non-virtualized nodes as well.

A VNF based on the implementation scenario can be deployed on single or multiple VMs as well.

Element Management nodes are responsible for typical management of one or several VNFs such as performance reporting, fault and event management, and so on.

VNF Manager

A VNF Manager is responsible for VNF lifecycle management (e.g. instantiation, update, query, scaling, and termination). Multiple VNF Managers may be deployed; a VNF Manager may be deployed for each VNF, or a VNF Manager may serve multiple VNFs.

Service, VNF and Infrastructure Description

This data-set provides information regarding the VNF deployment template, VNF Forwarding Graph, service-related information, and NFV infrastructure information models. This template information is used by the NFV-MANO layer and used for functional blocks management and orchestration.

Operations Support Systems and Business Support Systems

The last layer is related to the operation and business support system of an operator and the NFV Orchestrator is in charge of orchestration and management of virtualized software resources.  

For the last word, hereinbelow, you can find the complete 8 years of work in NFV ETSI ISG. BTW, I have to mention that one of the new concepts, which is the choice for virtualization of the network in the future, is cloud-native. Cloud-native inherit some specifications from NFV and is on the progress.

 https://docbox.etsi.org/isg/nfv/open/Publications_pdf/Specs-Reports

Many thanks