Requirements for Multi-Service and Multi-Tenant Access
In the optical access network (OAN), optical distribution network (ODN) is a key resource and an important asset for operators. In addition to vigorously deploying FTTH-based home broadband services in the areas covered by existing ODNs, operators also actively develop other services and explore new business models including network leasing/sharing to boost revenue and shorten the investment return cycle.
In a multi-service scenario, the access network concurrently provides access to multiple services. In other words, multiple services share the same access network. An operator uses the same access network to operate multiple services and offer differentiated QoS guarantees. Typical fixed-line multi-service scenarios include home broadband as well as government and enterprise business. Typical 5G multi-service scenarios are enhanced mobile broadband (eMBB), massive machine-type communication (mMTC), and ultra-reliable low latency communication (uRLLC).
Multi-tenant sharing allows an access network to simultaneously provide services for multiple operators, that is, multiple operators share the same access network. The operator who owns an access network is called the infrastructure provider (InP), while the operator who leases the access network for business operation is called the virtual network operator (VNO). As agreed in the business contract, the InP logically partitions the access network and assigns the resources of a logical partition to a VNO so that the VNO can independently operate them.
Multi-service and multi-tenant access can be understood as two dimensions of access network sharing. Multi-service access is the horizontal dimension, where the access network is shared by different services; multi-tenant access is the vertical dimension, where the access network is shared by different operators. The two dimensions can exist concurrently. For example, a public access network is planned in the Xiong’an New Area in China’s northern province of Hebei. The access network can be shared by the three major operators in the country—China Unicom, China Mobile and China Telecom. Each operator can operate multiple services such as home broadband, base station backhaul, and enterprise leased line.
In a complex access network sharing scenario where multiple services and tenants co-exist, different VNOs provide their users with services that fulfill respective service level agreements (SLAs). The access network needs to support concurrent, independent management by multiple VNOs and to provide differentiated QoS guarantees for multiple services. To implement multi-service access and multi-tenant sharing of the optical access network, an InP must consider and plan for network architecture evolution and select a proper new-generation optical access solution.
Introduction of Network Functions Virtualization
Network functions virtualization (NFV) has the ability to provide network function as a service (NFaaS). Therefore, NFV can be introduced in the access network to meet the need of network functions as different roles such as end users, VNOs and an InP in a complex multi-service multi-tenant scenario.
Currently most network devices are purpose-built in their entirety or their hardware. They come from only a few vendors, cost a lot and are difficult to scale. These problems can all be addressed by NFV. Through functional analysis, logical partitioning, network function (NF) module setting, NF module function definition, NF module interface definition and NF module development, NFV enables one or more virtual network functions (VNFs) to run on commercial off-the-shelf (COTS) hardware. Thanks to NFV, a combination of COTS hardware and VNFs can even be used to replace some types of purpose-built network devices and hardware.
NFV decouples network functions and purpose-built devices, offering operators more choices and greater flexibility in the use of network equipment. In addition to purchasing COTS hardware like blade servers and storage devices, operators can even develop VNF software to implement specific service management and control functions to provide differentiated services.
To enable application scenarios where the access network is shared by multiple services and tenants, the new-generation optical access solution needs to have the following capabilities:
● On-demand network connectivity that allows for easy control of the connections between end users and VNFs
● VNF as a service that facilitates customization by VNOs
● NFV infrastructure (NFVI) that enables VNOs to develop their own VNFs.
TITAN Enables Multi-Service and Multi-Tenant Scenario
ZTE's TITAN is a new-generation OLT that supports NFV evolution and can meet the need for multi-service and multi-tenant applications in an access network sharing scenario.
The NFV deployment strategy of the TITAN solution complies with the BBF TR-384 specification—Cloud Central Office Reference Architectural Framework
. Some highly real-time service management and control functions, including PLOAM and DBA of GPON as well as LACP, xSTP and OAM of Ethernet, are encapsulated as physical network functions (PNFs) and kept in TITAN. The other management and control functions can be deployed according to customer requirements either in VNFs in the cloud or in PNFs in physical devices (Fig. 1).
A highlight of the TITAN solution being applied in an access network sharing scenario is the implementation of physical resource abstraction, data modeling and resource mapping on generic compute and storage resources in the cloud. This implementation is carried out by ZTE’s ElasticNet unified management expert (UME). UME runs on the cloud and consists of two layers. The southbound or lower layer maps physical resources, while the northbound or upper layer abstracts virtual resources.
In its physical resource mapping layer, UME sets up a 1:1 physical access node (pAN) for each physical OLT. By using an independent database to store the static configuration and dynamic status information of every pAN, UME supports the offline configuration of physical access devices. Multiple physical resource management and control module instances can be created in the physical resource mapping layer of UME. Each instance can concurrently manage and control multiple and various pAN entities. For example, for different OLT models (C610, C650, C600, etc.), their software and hardware characteristics are represented in corresponding pAN data models. In addition, UME supports multiple pANs that are connected by physical links and can be simulated into one pAN to implement cluster management and control based on multiple-to-one mapping.
In its virtual resource abstraction layer, UME extracts pAN instances related to the data model of access-network-sharing services. Such instances include logical Ethernet subinterfaces, forwarding instances, traffic management (TM) characteristics (scheduling and rate limiting), performance management (PM) characteristics (statistics), and alarms. Meanwhile, UME ignores the information or characteristic irrelevant to the data model of access-network-sharing services, such as power supplies, fans and line cards. By combining resources as needed, UME creates a logical access device entity, establishes a data mapping link, and sets up a corresponding virtual access node (vAN). Resource combination can be based on a service, a VNO, or a service of a VNO. UME also uses an independent database to store every vAN data and supports the offline configuration of logical access devices. Compared with the pAN data model, the vAN data model focuses more on service description. For example, vAN in an OLT usually adopts an L2 Ethernet switching equipment model, while related characteristics of L1 interfaces (such as PON ports) are kept in pAN.
pAN is an abstraction of a physical device while vAN is an abstraction of a logical device. When UME processes the two layers of abstraction, it supports capability adaptation and coupling between pAN and vAN. The resource isolation level supported by physical devices will affect the configuration mode of vAN. For example, ZTE’s OLT C300 supports isolation between forwarding instances including VLAN and VRF but not between the packet buffer and MAC address table. In a UME, C300 pAN will display this isolation capability. When UME manages multiple vANs in the same C300 pAN, it will use the isolation capability and employ techniques like address isolation/arbitration to prevent logical resource conflicts between vANs.
The TITAN-enabled NFV solution has the following characteristics:
● The virtual resource abstraction layer of UME provides stakeholders like VNOs with programming interfaces for virtual access devices.
● The physical resource mapping layer of UME offers management and control interfaces for physical access devices. The plug-ins of a third party can be used to manage access devices of other vendors.
● UME supports several mapping modes between physical and logical devices, such as one-to-multiple slicing mode (where one physical device is partitioned into multiple logical devices) and multiple-to-one expansion mode (where multiple physical devices are merged into one logical device).
● The deployment of UME is based on an NFV architecture that uses off-the-shelf compute and storage resources. UME supports virtualization technologies including virtual machines and containers. Microservices are leveraged to enable independent evolution and flexible expansion of different management and control functions as well as different instances in the UME.
ZTE's NFV solution in the access network opens up a path of evolution from a single FTTH scenario to multi-service and multi-tenant scenarios. With the solution, end users, VNOs and InPs can define their own VNFs and build their respective virtual access networks according to their different roles in the access network. The access network infrastructure will ultimately be shared by multiple services and multiple tenants.
Multi-service and multi-tenant scenario, NFV, Access network, TITAN, virtual access node