 | Issue: | Europe II 2014 |
| Article no.: | 10 | |
| Topic: | The future of SDN is certain – but the path to it remains unclear | |
| Author: | Stefano Pileri | |
| Title: | CEO & co-founder | |
| Organisation: | Italtel | |
| PDF size: | 217KB |
About author
Stefano Pileri is Italtel’s CEO. Stefano Pileri, born in Rome in 1955, is an Electronic Engineering graduate and holds a Masters degree in Applied Electromagnetics. He took the position of Italtel Chief Executive Officer in September 2010, tasked with leading the company through financial restructuring, following a new industrial strategy.
Before joining Italtel, Pileri held various positions at Telecom Italia, including Head of the Network, Chief Technology Officer and Director of Technology & Operations. He began his career with SIP in 1982. From 2009 to 2011 he served as President of the Federation of Confindustria Innovative and Technological Services. Between 2007 and 2010, he served as Vice President of the Union of Industries of Rome, taking on responsibility for the Digital City of Rome project.
Pileri has received several international awards, including the Tele Management Forum Award, the International Engineering Consortium Award in the USA and EUCIP Champion.
Article abstract
SDN sounds like it is too good to be true – and it is, for now. It lacks standards that will ensure interoperability, so service providers are reluctant to commit to its implementation. However, the standards are already on their way, supported by the major players in the industry. Rolling out SDN will still require complex system integration, where independent system integrator can assist, especially with multi-vendor open solutions. Through precision SLAs, on-tap QoS and efficient resource management, service providers can claw back revenues that have been drained away by OTT up till now.
Full Article
First conceived from work undertaken at the University of California and Stanford University in 2008, Software Defined Networking (SDN) is still new enough to be classed as an emerging technology and therefore is on the radars of large companies less than other more established solutions. However, unlike some other experimental concepts which ultimately fail to live up to industry expectations, SDN’s promise to revolutionize networking is fully achievable, making it the technology that everyone, whether they are network operators, vendors, service providers or platform developers, should be putting on their agendas for 2014.
The need for SDN has risen from operators’ increasing need to simplify and reduce the cost of infrastructure management by abstracting network control and making the underlying layer ‘dumb’. Other factors include: the explosion of Big Data, The Cloud, the Internet of Things, Machine-to-Machine, multi-media entertainment, mobile traffic and a host of other established and emerging trends. Present IP Network infrastructures are not flexible enough to cope with all of these trends, both in terms of resource allocation and network path optimization, and as a result are becoming congested, affecting quality of service.
Why SDN?
In order to adapt to the situation, the network needs to become more application aware and should be able to communicate the needs of applications to the network infrastructure. With SDN, together with network programmability functions, communication between the network and upper-layer applications becomes bidirectional. Consequently, transparent traffic flows can be managed across the entire network, without specific knowledge of switching and routing protocols. Resource usage is always under control and infrastructure management becomes easy and cost-effective.
The advantages that SDN brings include integration and simplification of the network, making networks programmable and more agile, while also creating opportunities for policy-driven monitoring and more automation. Simply put, SDN enables networks to keep up with the speed of change. This is made possible by the virtualization of resources, making it the perfect match for cloud computing.
SDN: The Challenges
If, at this point, SDN sounds too good to be true, it is because it currently is. SDN in its current state continues to present numerous challenges to the telco industry and there is a lot of work to be completed before SDN can realize its full potential. Perhaps one limitation in the development of SDN is the current lack of standardization. For many years, networking has enjoyed standard protocols, as compliance with these standards means interoperability. This common vision is not currently present in the many components that make up an SDN network. As a result, SDN adoption is slower than it could be, while operators remain reluctant to move forward and commit to new technology without the guidelines that they have grown used to.
The first area where the lack of SDN standardization is evident is the range of managed capabilities that are available for SDN control. While there is currently a basic set of capabilities, each vendor tends to also state very specific capabilities which they believe have additional value. Similarly, there is no standardization of hardware in SDN and vendors have to deal with a variety of different procedures, keeping SDN very much at the beginning of its journey, at least until a common vision is agreed.
Agreement is also required on the timeframe for SDN, when should platform developers, network operators, service providers and vendors be doing what. Service providers need to decide if they are going to use SDN for certain applications, therefore overlaying the current network with SDN, or whether they are going to gradually include SDN as a potential interface within their current network. This is important because there are some incumbent players who will promote their brand for their routers to expose SDN interfaces in order to achieve SDN promises. This is in contrast with other vendors who are proposing vendor switches with are compliant with Openflow controllers and centralized IP routing. The danger here is that if this continues, SDN will grow organically and become fragmented by individual interests and domain specific implementations.
For a real SDN deployment on service providers’ networks, it is necessary to develop use cases that allow aspects of the network, for example, Operations, to be simplified and therefore reduce OPEX. It is not likely that operators will introduce SDN on their entire networks all at once. Hence, by identifying concrete cases of application, implementation of SDN based on vertical solutions from different vendors may go ahead.
Overcoming the Obstacles
Despite the challenges, realizing the SDN potential remains entirely possible and we are confident this will happen. The most important thing to enable SDN to deliver its promises is for platform developers, service providers, network operators and vendors to discuss interoperability. The reason we can be confident that SDN will be a success story is that this conversation has already begun.
Network operators, who incur the cost of operation and maintenance, need to discuss what impact the introduction of SDN will have on the existing technology, because, as always, there will be implications. With the network inevitably becoming application-centric, application developers need to take the opportunity to express what they require from the service provider in the underlying infrastructure and in terms of latency. For the service providers, it is important to break the ‘wait and see’ stance and decide how to move forward.
The cause for optimism comes from the work which some organizations are already carrying out. OpenDayLight (ODL), an open source project of the Linux Foundation, accepts software contributions from vendors and individuals alike, with the aim of creating an open source SDN stack that defines the controller and the interfaces. This is a significant effort, supported by Cisco, Hewlett-Packard, Microsoft, NEC, Red Hat, IBM, Juniper, Arista, Brocade and VMware and others. The project has progressed quickly due to the support of major players, defining features up and down the SDN stack, which are well positioned to be accepted by all SDN vendors. Vendors will be able to sell an ODL-compliant product, while also offering enhanced functionality as add-ons. While there is still some way to go, particularly in terms of awareness among IT staff, ODL gives an example of what could become a de facto standard for SDN.
Taking all of this into account, one of the other most fundamental requirements to optimize SDN is to have a system integrator who understands the network and how the applications run. SDN implementations must be aware of the whole network with all its components and services, and avoid localized point solutions. This is where an independent system integrator can help. Besides providing third-party technologies for the realization of the network infrastructure, system integrators combines the best innovative solutions to enable SDN in a multi-vendor approach. Some system integrators also design and develop proprietary products to enhance the structured communication system. Our software applications, for example, work in a virtualized environment, interfaced with SDN controllers and use IMS and OTT technologies, therefore they allow operators to exploit their assets, serve their applications’ needs and provide service differentiation.
What SDN means to service providers and end users
Perhaps the most important thing SDN can do for service providers, in addition to increased control, is to provide them with the opportunity to reduce operating costs and generate additional income.
The explosion of OTT applications means that service providers have been losing out on application revenues as customers move from the traditional means of communication to free internet applications. SDN gives service providers the chance to claim some of this market back. Web RTC, which enables audio, video and data services through the implementation of open-source APIs and Java scripts, is one of the most effective methods of doing this.
Another relatively simple way service providers will be able to increase their income is through the implementation of more robust Service Level Agreements (SLAs) which can be customized to a particular segment, such as stockbrokers, who require a certain level of service at a certain time of day. SDN makes this possible, due to the increased flexibility and the precision, allowing service providers to be very specific about what end users can expect and when. This, in turn, could lead to reduced service cost for the end user.
Conclusion
There is little doubt that SDN holds a great potential to transform the telco industry. Although there are serious challenges, the general consensus is that it is worth doing, and worth doing now. There is already much work being done to help sorting out exactly what SDN is and what it can do, and I’m confident that within the next twelve months we will see SDN growing in both the number and functionality of network applications that it supports, eventually transforming the network as a whole.
