Darryl Edwards Issue: Europe II 2014
Article no.: 3
Topic: SDN in Our World Today
Author: Darryl Edwards
Title: President & CEO
Organisation: ECI Telecom
PDF size: 275KB

About author

Darryl Edwards is President and CEO of ECI Telecom. With more than 30 years of experience in the telecom industry, he was most recently the Chairman for MACH, and telecom advisor to Warburg Pincus. He is a non-executive director of EXFO Inc. (NASDAQ: EXFO). Previously, he was CEO of AIRCOM International. Darryl spent 17 years at Nortel Networks in several executive positions, including Chairman of the Board for Nortel’s interests in Turkey and the Nortel’s Israel Joint Venture, CEO of Nortel Germany, President of EMEA and President of Global Sales. Darryl began his career in GEC Telecommunications (later GPT). Previously, he was a member of the Advisory Council to the Turkish government and also served on the UK Government Broadband Stakeholders Group and the Information Age Partnership.
Darryl Edwards holds a Higher National Certificate (Physics) from Birmingham Polytechnic in the UK.

Article abstract

SDN separates infrastructure from the control plane, and provides APIs to the application and management layers. Carrier-Grade SDN is more demanding than what was conceived for data centres. It needs to interoperate between networks, to manage different equipment including legacy, and to ensure QoS and high availability. Carrier networks are affected by SDN Transport, with resource optimization to support bursty demand and cyclic pattern network loads. SDN Management provides easier, faster provisioning and better utilization of existing capacity, even serving dynamic SLAs. Application aware SDN, such as event calendar, allows price differentiation for reserved capacity for a few hours or other types of bandwidth-on-demand and scheduled service resourcing. SDN Network analytics provides a view of the whole network, enabling better decisions to be made. 

Full Article

There is no debate that the telecom sector is once again undergoing dramatic changes. Service providers’ challenges seem to repeat themselves, as demand for bandwidth continues to grow – every day, more than 15 petabytes of new information is being generated. Cloud-based services are responsible for much of the communications traffic today, as individuals make frequent, massive and intensive use of the cloud, for services and applications such as Google, YouTube, facebook or whatsapp. Organizations and enterprises of all sizes are increasingly basing their communications infrastructure on cloud services as well.
To make matters worse, network providers are unable to raise prices, as competition continues unabated, both from other infrastructure providers and from over-the-top content providers. What’s a service provider to do? Obviously, doing the same as ever – i.e., buying more equipment and upgrading your network to achieve more bandwidth capacity – will take you only so far.
It’s time to get smart about it. The idea is to dramatically improve bandwidth efficiency of deployed equipment, and also realize that new equipment will further enhance the bandwidth/cost ratio we are used to.
Welcome to the world of software-defined networks, the-much-talked-about SDN (Software-defined networks).

The promise of SDN
For the past year, I haven’t been in one meeting or seen one presentation that did not mention SDN, or a variation thereof. SDN brings with it the potential to change the dynamics of the telecom business today. Once upon a time, we looked at networks as being somehow predictable and static, offering few and simple services and implementing changes only when absolutely necessary. The past five to ten years brought progress in the form of statistic and dynamic configurations, multi-services, and the ability to change things a bit more efficiently. With SDN, our networks will go even further, becoming increasingly programmable and flexible, and enabling use-as-you-need capabilities, multi-applications, and real-time changes.
SDN, as the name implies, puts the software in the centre of your network, as it befits the critical role it plays. The concept of SDN is to separate the network element – the infrastructure – from the control plane, where changes can be implemented more efficiently and cost-effectively. Equipment deployed in the field will be common (and therefore inexpensive) units, offering multivendor interoperability, which will obey precisely the instructions of the central control plane. SDN provides APIs for upper layers, which means applications can utilize the network, request available resources, or assign resources dynamically, all according to their needs at any given moment.
SDN brings three main benefits to service providers:
1) Creation of new revenue streams, because these networks are open, programmable, and application-aware, which allows them to differentiate services, both in price and priority, according to service level agreements (SLA.)
2) Lower capex, because SDN allows better network utilization and multi-layer scalability, without adding or replacing deployed network elements. This is not only true for new equipment, but also for deployed units, which will be able to significantly improve their efficiency once traffic is managed by a central control.
3) Reduced opex, because SDN enables faster upgrade time for both hardware and software via centralized operation, and it saves energy by improving equipment utilization.
What’s not to like?
Too good to be true?

It’s easy to get carried away with one more buzzword, so we should remember that SDN was originally developed for data centres and enterprise usage. It is no surprise that our customers – service providers and utility operators – look at SDN with high expectations, but also make sure to ask many questions. After all, unlike data centres and enterprises, service provider networks are characterized by:
• The need for high availability, reliability and quality of services
• The need to interface and interop with other networks
• The need to integrate different technologies, different platforms (of same or different vendors) and diverse generations
• The reliance on a distributed infrastructure, rather than a centralized cloud.
No question that SDN is a disruptive architecture, with potential deep impacts in the whole telecommunications industry. However, we would do well to remember that SDN is still in its infancy and there are still many challenges and ‘open’ issues, such as interoperability not only among vendors, but also between legacy and next-gen equipment; network redundancy, robustness and control protection; and reliability and low latency as expected by end users.

We are not ignoring any of these issues. In fact, the sharpest minds in the industry are discussing these very topics in special committees set up by the various regulatory agencies. There are no final answers yet, but I am confident that it is just a matter of time.

How can SDN benefit telecom networks?
Even with the open issues and concerns, the promise of SDN brings remarkable benefits to our industry. After all, the main challenges facing our customers today are how to enable economic and efficient resource allocation to support bursty demand for bandwidth caused by the cloud computing/networking phenomenon; how to keep pace with developing networking standards and protocols while protecting investments in already-deployed equipment; and how to establish new services or to provide access to brand-new applications that consume network services, without replacing or requiring extensive integration with existing network equipment.
SDN can address these challenges, as it provides our customers with an intelligent network that is more automated. In addition, by applying SDN at the transport layer of the network, we accelerate service provisioning and differentiation without the need to re-architect the network every time. Lastly, it guarantees the same level of network control our customers expect and are used to today.
SDN at the transport level takes into account the parameters resulting from the constraints of each technology used in each section of the network. However, transport-SDN is not just a matter of protocols and interfaces, but is primarily a holistic network analysis for real-time optimal traffic steering and resource allocation to meet applications’ requirements on-demand.
Some benefits from application-aware transport
An application-aware transport network allows dynamic resource allocation and multilayer/cross-layer on-the-fly optimization to support flexible and agile service creation, better network utilization, and scalability without adding/replacing their already deployed network elements.
Application-aware transport networks are a long-term remedy for cost-effectiveness in service providers’ networks, as they allocate resources when, where and as long as needed. Furthermore, they allow cost-effective bandwidth-on-demand and scheduled service capabilities.
The keyword here is flexibility.
With bandwidth-on-demand, both customers and external business applications can allocate network resources easily and dynamically, to increase or decrease resources assigned to applications according to actual needs, and to use services only when they are really required. Bandwidth-on-demand saves network resources and dramatically improves network utilization as well as network elasticity.
Similarly, as many services are required only at certain times and for predictable durations, service providers need a way to schedule resources’ allocation according to an exact timeline. For example, SDN helps operators manage cyclic pattern network loads: during the daylight hours, most of the traffic is generated by trade and industry users, while there is far less need from residential areas. In the evening, by contrast, most of the traffic is domestic, while the industrial traffic declines (Fig. 1.)

(insert) Fig 1: Cyclic schedule of services

Another predictable example is more ad hoc in nature, but can cause serious network congestion: A major sporting event is held in a municipal stadium with a huge amount of people concentrated in one place for several hours, who are now sharing photos and video clips using their smartphones, and unwittingly causing massive bandwidth and network service overload. Such an event endangers network balancing efforts, due to the sudden deviation from normal traffic patterns.
Business applications and service operators both can benefit from a calendar-based solution to reserve resources in advance, allowing explicit scheduling of a network service (start time, end time, involved end-points, required bandwidth) for a specific amount of time.
SDN and application-aware transport are also helpful in avoiding network traffic congestion and bottlenecks that can cause service interruption. To ensure enhanced operation and customer satisfaction, service providers essentially need a congestion relief mechanism that keeps track of the actual network traffic and anticipates how certain events can adversely affect the quality of service. When such a situation is detected, a proper congestion relief mechanism is activated, diverting traffic to where there is available capacity. Congestion relief saves costs and improves network utilization (Fig2.)

(insert) Fig 2: Dynamic congestion relief

Building the Ideal SDN Network
The ideal SDN solution features the required openness that is essential to the SDN concept, in addition to the programmability and automation necessary to make it ‘application-friendly.’
SDN’s greatest advantage lies on its ability to completely control equipment deployed in the field, from a central location. From one (virtual) place, operators have total control of their networks, better utilizing resources, reducing costs, and reaping more revenue. SDN brings the holistic network analysis and control required for application-aware transport to allow optimal traffic steering and better resource allocation, according to applications’ requirements on-demand. SDN allows our customers to enjoy a larger share of the telecom revenue pie, while dramatically decreasing their capital and operational expenses.
It is exciting times indeed, when we can live to see such major changes in how we build, operate and manage our telecom networks.

A number of experimental operational networks are being installed in Europe. One case that has been publicized is Deutsche Telekom’s TeraStream project in Croatia http://www.slideshare.net/ceobroadband/dt-tera-stream. Colt also announced its SDN-Based Carrier Ethernet network , allowing for higher bandwidth delivery, more automation and scalable business Ethernet services http://de.slideshare.net/ColtTechnologyServices/colt-data-centre-and-network-programmability-sdn-summit-2013