Etisalat — 2020 Landscape
Two major disruptive technology innovations are shaping the future of telecommunications — Software-Defined Defined Networking (SDN) and Network Functions Virtualization (NFV). Further, emerging services based on super-rich content and the Internet of Things (IoT) have stringent requirements for ultra-low latency, high bandwidth, seamless mobility, and personalized subscription efficiency. Each of these imperatives has shaped the ‘cloud’ as the de-facto technology for future telecom networks.
Fundamental changes in future networks will coincide with a tailored, road-mapped formulation of SDN/NFV deployments that can help achieve the following objectives:
- Exceptional Customer Experience: The provision of online, on-demand, and real-time experiences across various channels and touch points for all new and legacy services
- Efficient Operations: 1) Simplified, agile operations for reducing Total Cost of Ownership (TCO); 2) faster deployments and improved scalability
- New Revenue Streams: 1) Monetize new architecture with open APIs; 2) use virtualization as a tool to offer customers new products and services; and 3) reach new customers and markets through the cloud
A distributed Data Center (DC) architecture will be at the core of the future network, and the SDN and NFV-driven transformation will eventually affect all key aspects of networks and operations. Software and hardware decoupling are mandatory, and virtualization will enable shorter deployments and faster times to market.
Transforming the End-to-End (E2E) network architecture will require an agile and cloud-centric approach.
The ambitious virtualization of existing network functions is essential — whatever can be virtualized will be virtualized. Though certain access network functions may remain ‘physical’ or take longer to be virtualized, future networks must be open, agile, and flexible enough to adapt to any new network access medium that may be deployed in the last mile.
2020 Network Concept
Etisalat’s goal is to achieve a 60 percent virtualization of network functions and 70 percent reduction in time-to-market lifecycles for all product and service developments by 2020. This strategy is driven by the rise in customer growth requirements in emerging business markets.
Figure 1: Overview of 2020 Goals
2020 Services Outlook
Future services will be bundled into three primary categories:
- Communication Services: Exponential growth in bandwidth and connected devices will remain a major factor in the Middle East, Asia, and Africa.
- Content+: The core of future digital services and business models includes rich content.
- IoT Services: Etisalat will bring pre-packaged innovations to a wide range of industry segments.
In all scenarios, the focus is on the development of new partnerships that leverage revenue from emerging ecosystems while deploying state-of-the-art technology for delivering best-case customer experiences.
Network Service Prerequisites
The popularity of Internet-enabled smart devices challenges telecom providers to eliminate network inefficiencies and offer agile, on-demand, cloud-based Internet services. This network concept will eventually allow Etisalat to extend its portfolio beyond connectivity to become a key player throughout the ICT value chain.
Enhancements in access technologies will result in a shift in the ratio of access, aggregation, and core bandwidths. The aggregation layer will become bulkier due to increased bandwidth and new ultra-low latency technologies, the result of which will be important shifts in content localization by 2020.
User experience will also play a major role in enhancing end-user bandwidth. The entire network architecture must be carefully planned to:
- Increase aggregation and core bandwidth
- Reduce the number of hops to reach the content
- Reduce the latency of access to key services
The adoption of IT technologies to enable the rapid launch of new services is another critical design criteria. Traditional infrastructures have gradually transformed to become adaptive ‘agile core’ infrastructures in order to quickly respond to customer demands for new services, as illustrated below.
Three key pillars of our future network architecture emerge from this concept: 1) Software-driven, data-center-based platform; 2) virtualized network functions and applications; and 3) agile and orchestrated operations.
Carriers of the future must be able to easily increase capacity, efficiency, and agility, cut TCO, and mandate an automated, scalable, virtualized, and open platform. The platform will focus on the synergy of three primary areas: 1) A connectivity architecture, where underlying IP and optical networks enable delivery of future services; 2) the full potential use of SDN and NFV technologies; and 3) initial focus on deployment of a scalable, open, cloud infrastructure.
Connectivity Reference Architecture
The evolution of the connectivity layer towards cloudification and virtualization requires understanding the evolving roles of each domain in the hierarchy of a carrier infrastructure.
Agile technologies add flexibility and speed to the core network. Regional Points of Presence (PoPs) and central offices are becoming the Cloud Edge. Regional PoPs are converted to a DC-based infrastructure with the dedicated hardware appliances and network functions formerly located in central offices that are now virtualized. Agile Core is the key enabler for the scalability necessary to interconnect external clouds so that vertical solutions can be seamlessly integrated for E2E services.
The Cloud Core can host Virtual Network Services (VNS), such as the Value Added Services (VAS), Orchestrator, Operations Support System (OSS), Business Support System (BSS), and analytical management networks. Interconnecting the Cloud Core with the Cloud Edge is the Connectivity Layer, which consists of the IP and optical environments. Etisalat will intelligently leverage IP and optical bandwidth with SDN and NFV applications that increase efficiency without compromising redundancy or quality.
Figure 2: Etisalat Planned Cloud Services Ecosystem
The Cloud Edge layer, hosting most Virtual Network Functions (VNFs), will include multiple access technologies that terminate on a Virtual Edge platform on which numerous other functions will also be virtualized.
The traditional access network is becoming an Agile Access Layer able to virtualize many types of high-speed physical technologies such as Gigabit Passive Optical Network (GPON) and copper-based G.fast.
Agile Metro Aggregation Layer
The Metro Network will also become agile to provide both better service quality and user experience. Traditional bandwidth availability at the aggregation layer ranges from 1 to 10 Gbit/s. By 2020, the bandwidth requirement at this layer will range from 100 to 400 Gbit/s due to the rapid growth in connected devices and the flattening of traditional telco layers.
Bandwidth availability at the core layer has traditionally been provisioned in multiples of 10 Gbit/s. Due to centralized and virtualized network functions, bandwidth-intensive applications and content requirements will surge to reach a range from 400 Gbit/s to 1 Tbit/s by 2020.
For Over-The-Top (OTT) providers, content localization will save significant bandwidth; however, external OTT providers using telco backbone infrastructure will require three times more core bandwidth.
Etisalat expects the evolution of SDN- and NFV-based technologies to reach the efficiency necessary to manage performance, security, and agility at this massive scale.
Despite growth in core network and Internet backbone traffic, the access layer is perceived as a source of congestion. Traditionally, the probability of congestion was higher in the access layer and the edge network. Advanced access architectures based on SDN and NFV are being designed to aggregate multi-gigabit circuits to better offload the choke points into the core networks.
SDN will provide automated control and a proactive capacity planning approach to improve backbone efficiency and productivity, though it is unlikely that any software technology will fully virtualize the core network.
SDN-enabled Network Domains
Though centralized resources are known to reduce financial and operational overhead, reducing latency in carrier environments requires that domain-specific controllers be deployed in a distributed fashion.
SDN deployments will use distributed controllers hosted at the Cloud Edge to keep with the general architectural goals for balancing economic and technical constraints, overcoming scaling issues, and eliminating complex east-west cross-controller communications between domains.
The SDN controller will compute control plane information and program individual network elements using flow rules, and centralization of the control plane on the SDN controller will eliminate MPLS routing protocols.
The SDN controller abstracts the network for use by the orchestrator and other upper-layer applications such as traffic engineering, flow-based load balancing, discriminated traffic steering, bandwidth on demand, bandwidth calendaring, and others.
Challenges in Emerging Markets
The development of automated provisioning platforms for multi-vendor systems is hindered by the lack of involvement by the international standards bodies.
This SDN-driven transition may be the only way to achieve true service agility as it affords a unique opportunity for vendors and telco carriers to work together to develop a dependable orchestrator for automated provisioning.
With the evolution of current MPLS networks due to NFV, the boundary between networks and compute functions is fading. In fact, someday, virtual Provider Edge (PE) functions, such as those required for multi-tenancy, segmentation, or regulatory compliance, may run on Commercial Off-the-Shelf (COTS) hardware.
The power of SDN includes the opportunity to narrow or eliminate the bridge separating networks and applications, as found in traditional hop-by-hop, Quality-of-Service (QoS) deployments. Applications need a clear view of the network and which path is best to ensure QoS, and networks need a clear understanding of the resource requirements for each running application.
Security features have traditionally been deployed in layers with varying degrees of complexity. The implication of the security architecture in SDN and NFV environments is that these same functions will be controlled centrally while remaining activated at the edges — a methodology that can be expected to lower OPEX.
In addition to reductions in capitalization and operating costs are the benefits in the security domain following the transition to a SDN/NFV platform, which will include improved operational efficiency, enhanced policy enforcement, and the availability of value-added services that were previously cost prohibitive or not feasible.
Among the challenges for network security in the SDN/NFV world will be the collection of telemetry data for forensics purposes provided by Deep Packet Inspection (DPI) from many points in the network.
Carrier Grade KPIs: One of the primary challenges slowing the adoption of SDN and NFV technologies by telcos is the limited performance of general-purpose Information Technology (IT) systems when compared to the performance of dedicated Communications Technology (CT) hardware. Operators require carrier-grade systems in terms of availability, performance, and security. For example, millisecond latencies are necessary for delivering acceptable levels of service for many carrier applications, per contracted Service-Level Agreements.
Figure 3: SDN Maturity is Predicted to Occur in Three Phases
Realistic Expectations about SDN and NFV
In addition to bringing true agility, programmability, and efficiency to telecommunication networks, there is no doubt that SDN will introduce new types of complexity. Beyond the hype, it is necessary that a careful, pragmatic approach is required that details objectives and expectations.
Realistically, the minimal use of purpose-built equipment in favor of adopting COTS hardware will dramatically shorten equipment lifecycles — and while this could be viewed as a drawback, Etisalat accepts that technology is evolving ever quickly and that our access to future services depends on following the trend.
Cloud Infrastructure Reference Architecture
Elasticity is a main feature of the cloud. Data center resources are optimally allocated and assigned to particular services or tenants from a pool of compute, storage, and network resources. For multiple-cloud DCs, enterprise administrators have traditionally moved workloads manually, from highly utilized clusters to under-utilized pools. In carrier network environments, however, manual workload portability is impractical, so today, carriers are moving toward automating the allocation of multi-cloud DC resources to optimize their distributed architectures.
Evolving Industry Challenges for the Cloud
Transitioning from the traditional model to an NFV Infrastructure (NFVI) has many advantages and also many challenges:
Network simplicity (or not): NFVI poses architectural challenges, especially in multi-vendor environments. The interoperability of VNFs developed by different vendors may cause integration difficulties, performance challenges, and intermittent failures.
Mitigating risks requires: 1) A clear certification process; 2) a complete map of services to be provisioned, including hardware components, underlying architecture, sets of VNFs to be deployed, and integration points between each VNF; and 3) a thorough verification and testing process prior to any service deployment.
Cascading Effect: The growth of virtualization, service density, and operational efficiency also risks an increase in workflow complexity. Complex orchestration and automation create the potential for single failures in one network component to initiate a succession of failures to cascade throughout subsystems or service chains in ways that are hard to control or troubleshoot. To avoid such problems, operators must implement rigorous quality control protocols, deployment tests, and verification processes, especially early in the transition.
Shifts from traditional telco Communications Technology (CT) to enterprise-style Information Technology (IT) architectures: The need for a carrier infrastructure that combines CT features and levels of reliability with the flexibility of IT-based systems has never been greater. As this transformation grows and matures over the coming years, traditional enterprise IT platforms will not be expected to cope with the rigorous service delivery requirements of the telecom carriers. The anticipated result is that telcos transitioning to ICT architectures will face many refreshes and build iterations and customizations to blend, scale, and adapt existing service delivery platforms based on their new IT tools and technologies. Broad-guage technical expertise and solid project management skills are strong prerequisites for proceeding.
Benefits of Cloud Technology in Telco Networking
The benefits of cloud technology in the data center space are especially obvious for carrier networks. The SDN/NFV revolution will require a near-total replacement of all traditional networking devices to accommodate the new technology, and past upgrades are susceptible to unused spare capacity upon tapering demand. Deployment times, budgets, and customer experience will all be negatively affected absent a comprehensive plan. Completed virtualization solutions are organized to tightly couple the ability to scale cloud infrastructures up or down based on real-time demand.
(In Part 2, we will explore how the move to virtualization will translate to network functions and operations.)
Headquartered in Abu Dhabi, Etisalat provides innovative telecommunications solutions and services to 167 million subscribers in 18 countries across the Middle East, Asia, and Africa.
This white paper is prepared in collaboration with Huawei. For more information, please write to egwhitepaper@etisalat. ae
Copyright 2015 © Etisalat. All Rights Reserved