SDN and NFV: Faster Than the Future
Software Defined Networking (SDN) and Network Function Virtualization (NFV) are gathering momentum as telecom carriers struggle to manage rapid growth in data traffic. Recent discussions have focused on the ability of SDN and NFV to make network and service provisioning more agile and flexible; but do they fully address the challenge facing carriers today? Only by enabling real-time insight into what is happening in the network and how services are performing, will the true benefits of SDN and NFV be realized.
For many years now, carriers have been struggling with the growing volume, velocity and variety of data. This is a relentless growth curve with no immediate end in sight. The challenge has been to manage this situation profitably. The solution, many believe, lies in optimizing resource usage and accelerating the deployment of new and innovative services.
SDN and NFV have been identified as potential enablers of this solution. Based on Commercial-Off-The-Shelf (COTS) hardware, the immediate benefit is the expected cost reduction for both vendors and their carrier customers. But, even more importantly, it is the flexibility that virtualized network functions provide and the agility that centralized software programmable control enables that brings the real value.
But, do SDN and NFV address all aspects of the challenge that carriers face today? There is a vital ingredient in the formula for success, which is little discussed; on what basis will the agility and flexibility of SDN and NFV be called into action? What intelligence will be used to make decisions?
The answer to these questions is real-time insight. Only by enabling real-time insight into what is happening in the network, and how services are performing, will the true benefits of SDN and NFV be realized.
Building Flexibility and Agility into the Network
SDN and NFV have evolved independently, but share many common traits. In the telecom world, they are increasingly seen as complementary solutions to building flexibility and agility into the network while at the same time reducing cost and complexity.
SDN and NFV are expected to provision services quickly with as little dependency on the network as possible. With virtualized network functions and centralized switching and routing control, it is expected to be possible to literally move the bandwidth, processing and storage resources required to where they are most needed. Building this infrastructure on COTS hardware helps to reduce cost, while providing scalability and efficiency.
There is confidence this can be done because this is the way enterprise networks and data centers are operated today, albeit for processing and storage resources. Virtualization is now pervasive and supports a vast range of cloud services. Even SDN has origins in the enterprise and data center world as a means to reduce cost and improve efficiencies.
Telecom carriers and vendors have seen the benefits that these approaches have delivered, not least of which is the speed at which new services can be deployed. It is the flexibility and agility that virtualization, centralized switching and routing control promise that has captured attention.
Agility and Flexibility: Only Half of the Formula
SDN and NFV, if fully realized, can provide the benefits that carriers are seeking with regard to provisioning services in a faster and more efficient manner. However, this is only one half of the formula for success. The key questions that need to be asked. “What will trigger changes to service and network configurations?” “When will the capabilities of SDN and NFV be called into action?” “For what purpose?”
The answer to these questions forms the other half of the formula. The full benefits of SDN and NFV cannot be realized unless these questions are understood and answered.
For many readers with telecom experience, the questions raised above might seem confusing. The answer for many would seem obvious; telecom carriers will plan and create services and then use SDN and NFV to provision them.
The problem is this ignores the fact that the telecom world is undergoing a massive paradigm shift that challenges deeply held beliefs on the role of carriers as service providers. The bottom line is that carriers are no longer in control of the services that are offered on their networks.
Once we accept this fact, we begin to realize that the obvious answer is no longer so obvious. The challenge that carriers are facing is not only provisioning services, but also adapting to the behavior of subscribers consuming the carriers’ and other provider services.
Rethinking the Illusion of Control
The underlying service being offered by carriers is Internet connectivity, but it is the hope and vision of many of them to provide more advanced services that can potentially compete with other content providers, often termed “Over-The-Top” (OTT) providers. These OTT providers are increasingly driving mobile data growth.
According to the Cisco VNI Index released in February 2014, video traffic was 53 percent of mobile traffic in 2013 and will be over two-thirds by 2018. Sandvine reports in the Global Internet Phenomena Report 2H2013 that Netflix and YouTube now account for more than 50 percent of Internet traffic on fixed networks in North America.
There is a wealth of innovation that has been enabled by allowing services to be offered over the Internet, whether they are cloud services or apps for mobile devices. But, the reality is that carriers, at best, will be one of a multitude of service providers running over the network.
Subscribers are becoming more mobile and using smartphones and tablets as their primary devices for connectivity, and more devices will be enabled to share data for machine-to-machine communication, such as cars, household appliances and even trash cans. In this environment it will be virtually impossible to control and predict how much bandwidth will be consumed, and when.
The key capability that needs to be established is the ability to quickly adapt to new situations. SDN and NFV are one half of the formula that enables fast re-provisioning. The other half is real-time insight into what is happening in the network.
Reacting and Adapting in Real Time
By monitoring what is happening in real time, it is possible to see what services are being used, by whom and how much bandwidth they are consuming. The technology to do this exists today and operates at speeds up to 100 Gbps.
It is also possible to capture this information to disk in real time. This allows a profile to be developed of normal network behavior over a given period of time. This profile can be used to establish thresholds or other triggers for detecting deteriorations in performance and quality, radical changes in consumption patterns or other anomalies.
By comparing real-time data on what is happening with the expected behavior established in the profile, it is at least possible to determine if something out of the ordinary has occurred. While this does not provide a prediction on data consumption, it does provide a powerful capability to adapt to a change in real time.
Linking Real-Time Insight to SDN and NFV
By monitoring network and service usage in real time and detecting anomalies, it is possible to react to unexpected situations. This is the front end to the management of SDN and NFV services. It provides the intelligence for SDN controllers and NFV management to act. This can be to reroute traffic, increase bandwidth in a certain area, change service parameters or even offer a new service or upgrade.
This approach also brings other benefits. The real-time information on network and service utilization provides insight into Quality of Experience (QoE). This is crucial for services offered by carriers, but is also a service that can be sold to content providers utilizing the carrier network, such as OTT providers. A similar service can be offered for security, as the approach outlined above can detect anomalies that can be correlated with information from security appliances to determine if an attack is underway. The ability to offer better quality and more secure connections should be attractive to OTT service providers and their subscribers.
Playing the Devil’s Advocate
Many might argue that this aspect of real-time monitoring should already be solved in the SDN switches or the hypervisor. For low speed networks, this will be possible. The challenge is implementing this capability in high-speed networks operating at 1, 10 or even 100 Gbps.
There are a broad range of appliances available today that analyze real-time data for management, test, security and optimization of networks and services. These are stand-alone, physical devices with extreme processing requirements.
One should remember that this kind of functionality was originally available in switches and routers, but it was determined that separate dedicated appliances were a better solution as the processing requirements were too large for routers and switches to accommodate. With the move to COTS hardware envisioned in SDN and NFV, it is hard to see that these requirements have diminished, especially as the volume, variety and velocity of data is increasing.
Another argument is that appliances can now be virtualized and run on virtual machines. This again is true for lower bandwidth applications, but at higher speeds, the virtual appliance will require significant processing resources that can threaten to consume CPU cycles that should be used by the other virtual clients the virtual appliance is supposed to be monitoring. This would seem to be counter-productive.
Therefore, there is a definite need for physical appliances that can analyze large amounts of data in real time. To work in a virtualized environment, they will need to be virtualization-aware with the capability to distinguish virtual LANs. With these key devices in place, the real-time insight into how the network and services are performing can be assured.
Assuring Agility, Flexibility and Adaptability
The technology for implementing real-time insight already exists and is ready to be deployed. Applying this technology to SDN and NFV management is not a technical hurdle, but more of a conceptual step. It involves a rethink of the role of carriers and accepting that there is no longer control over what services are being consumed on the network. This involves an acceptance that rather than being the sole cradle-to-grave provider of all communication service needs, a modern carrier needs to partner with a growing ecosystem of OTT players with a focus on maintaining customer satisfaction. In other words, ensuring that subscribers are getting what they want, the way they want it.
Once we accept the reality that carriers are no longer in control of the traffic that is running on their networks, we begin to realize that the promise of agility and flexibility needs to be complemented by the real-time capability to adapt. With real-time insight into network and service usage, triggers can be built in to bring SDN and NFV capabilities to bear in addressing issues and providing services. In addition, the real-time insight can be used to assure QoE and security that in turn can be sold as services to OTT service providers. Integrating this capability into SDN and NFV concepts will complete the formula as part of a holistic approach to building agility and responsiveness into carrier organizations