Virtualized networks, real problems: How NFV needs network intelligence to address QoE concernsSites - Connect World
Network virtualization is prevention rather than the cure. Getting access to both – and taking a tangible step towards alleviating QoE challenges for good – depends on having visibility into what’s happening on the network at any given time, which is a whole different ballgame in a virtualized environment.Read More
The networking game has changed, and the challenges faced by mobile and broadband operators have never been more complicated than they are right now. Today’s subscribers are repeatedly hitting the network with heavier data requests than before, fuelled with bigger expectations and a fundamental, hardwired, reliance on this technology to power their lives. Maintaining a consistent QoE in this environment has become the sword operators live or die by.
Yet it’s not a new threat to their revenues. This problem has been faced for some time, particularly with mobile operators where the move from 3G to 4G, and with 5G now a real ity on the horizon, has exacerbated this further. There’s no doubt this issue will continue to grow as it isn’t going away any time soon, and while steps have already been taken to address it there’s still a rocky road ahead. The reality is that maintaining QoE in line with growing subscriber expectations has never been easy. This new environment opera tors have found themselves in, however, where demands on the network are reaching all-time highs and new apps are launching almost daily to create spikes in subscriber traffic, has cre ated the perfect storm of performance and bandwidth issues. This, in turn, has opened the floodgates for new technologies to swoop in and present a potential answer to what’s become a longstanding concern. The answer, at present, comes in the form of network vir tualization via NFV and SDN. Yet while this approach certainly goes some way towards al leviating operator concerns in relation to QoE by giving them the tools to quickly scale and become more flexible, it’s not a magic bullet. Network virtualization is prevention rather than the cure. Getting access to both – and taking a tangible step towards alleviating QoE challenges for good – depends on having visibility into what’s happening on the network at any given time, which is a whole different ballgame in a virtualized environment.
Network virtualization: The current state of play
Few would deny that network virtualization is essential for the future of networks and as a vehicle for operators to remain competitive. It’s widely accepted that subscribers expect access to everything, at any time, and from any location, which presents a two-fold issue for operators; how do they deliver the new apps and services required to grow revenues, while simultaneously juggling the need to keep pace with capacity demands and overcome the inherent operational deficiencies associated with legacy systems? Network virtualization has an obvious role to play. It’s no surprise that NFV and SDN have become the architectures of choice for net works of all shapes and sizes – in theory, at least. The reality is the shift to network virtualization hasn’t been plain sailing, partly due to the performance challenges that operators arealready facing.
Although both SDN and NFV make it easier for operators to roll out new services, making changes to the network in minutes rather than days or weeks, until recently operational effi ciency has been a stumbling block. Although there’s clear potential for this technology to address network scalability concerns in the long-term and also reduce OPEX costs, data throughput requirements and performance demands are constantly growing and pose a press ing challenge right now. Initial virtualized net work instances couldn’t always keep up with this, which slowed the growth and ultimate rollout of virtualized technology.
Fortunately, many vendors are increasingly pushing the boundaries of what’s possible when it comes to virtualized network performance. In the past few years alone we’ve regularly seen industry headlines of more than 1Tb/ps of data throughput on a virtualized network. Just last Network Virtualisation month, Procera pushed the envelope even further by hitting over 1.8Tbp/s. Unsurprisingly, operators are now sitting up and paying attention to the reality of network virtualization.
The potential for future scalability of this technology has been realized, making it practical for virtualized networks to reach critical mass during 2017 (Survey sponsored by Ciena finds that more than 30 percent of Asian enterprises have already implemented software defined networking).
But this is only one piece of the puzzle when it comes to addressing the QoE challenges operators are currently facing. Although virtualized networks give them the scale, through put, and performance required to meet future subscriber demands, this doesn’t address the other aspect of keeping network users happy – identifying problems in real-time and addressing them before the subscriber ever becomes aware.
The extra complexity associated with virtualized networks, and the additional internet working challenges of this technology running on COTS equipment, is both a blessing and a curse. It’s threatening operator efforts to protect the subscriber experience, resulting in a far more difficult environment for identifying the applications behind high volumes of traffic that can cause all manner of issues for the network.
The other side to network virtualization
Tackling this much greater problem depends on a number of different factors. First and fore most, operators need end-to-end visibility into all new services and platforms likely to cause a surge in demand on their networks, alleviating concerns with poor network performance driving subscriber churn. It’s also important for operators to know when and where a network surge is likely to occur, which comes back to the pressing need to be able to identify specific applications on virtualized networks. After all, it’s only by having this powerful level of actionable data they will be able to start meeting the needs of today’s subscribers en masse rather than simply serving a small heavy data usage subset of their overall user base.
Network intelligence and traffic management tools, supported by Deep Packet Inspection technology, are therefore vital. Granted, this has always been the case, but if anything they’re now all the more important in a virtualized network environment. Not only is DPI essential for unlocking the data needed to make this a reality, it makes it possible for operators to see what data is flowing across their networks in real-time, to prioritize traffic where appropriate, and to identify where the network is congested in order to take steps towards addressing it. What’s changed in a virtualized environment, however, is how DPI technology needs to be deployed.
New networks, virtualized insights
The ultimate goal of any SDN or NFV environment is to deliver an infrastructure capable of being configured on the fly for dynamic service delivery. As mentioned, making this a reality depends on network operators having access to analytics to identify traffic in real-time. Yet DPI in its current, physical form can only go so far.
Increasing support and reliance on virtualized networking technologies presents its own hurdle that must be overcome as a result. Since virtualized networks can be highly distributed and can change their behavior and traffic routing protocols based on current network conditions, this presents further application identification concerns. COTS hardware presents its own challenges for packet processing too. Both of these trends fundamentally change how DPI engines need to operate in new operator network environments in order to successfully deliver their core function of application identification.
In order to maintain the end-to-end visibility required for application identification to function, DPI tools must also become virtualized in order to have the same level of speed andefficiency as the virtualized networks they’re deployed on. But, even then, it’s not enough to simply restore network visibility. Although of fundamental importance, it’s equally essential for operators to partner with a DPI vendor that’s constantly updating their signatures database to ensure traffic can be accurately identified. Applications evolve quickly in the internet economy. A change made by a major service (Netflix, YouTube, Skype, etc) could cause huge inaccuracies in analytics or throw off QoE management if network traffic associated with that application wasn’t appropriately classified.
End-to-end virtualized intelligence
Unlocking end-to-end visibility into data traffic is only scratching the surface of what’s possible with DPI. But combining this technology with the flexibility and scale of virtualized networks has the potential to tip the scales back in the operator’s favor. Once the fundamental application identification challenges have been solved, several new opportunities open up. As has long been the case, speed is of paramount importance when it comes to QoE management. Taking the right approach to virtualized networking by putting the right tools in place to identify traffic in real-time makes this easier than ever before. Subscribers can be reached with real-time offers in response to their usage habits and adjustments can be made to network policies on the fly. Operators can act on this information to prioritize decisions around CAPEX and OPEX investments. The network visibility afforded by DPI also allows engineers to analyze usage on different parts of the network to manage future expansion, while marketing teams can differentiate against the competition for which services the operator is particularly strong in delivering.
With the increasing adoption of NFV, measuring the needs of new applications and providing a detailed assessment of a network can become a competitive differentiator. End-to-end visibility across a network allows operators to prioritize their investments into new technologies and expansion, achieving maximum ROI by identifying what changes will have the biggest positive impact on the overall subscriber experience.
Ultimately, on their own, virtualization technologies such as SDN and NFV aren’t a magic bullet for operators to tackle the QoE challenges they’re faced with. But combine the scale and performance benefits of this with analytics technology that can keep up, and operators have access to a host of new opportunities to keep subscribers happy, better manage network capacity issues, and access new revenue streams.