How to level the playing field with SDN
Waving an SDN (Software Defined Networking) magic wand over traditional network operator infrastructure to make it ‘just like OTT’ is a good deal easier said than done.
So says Ronan Kelly of ADTRAN, EMEA & APAC. OTT players deserve the credit for identifying and exploiting the new web-scale SDN principles that turned the whole telecoms market on its head. But when your infrastructure starting point has zero legacy, that’s a comparatively simple thing to achieve.
The answer can’t be to wait any longer for transition to SDN to become easier. A growing cohort of operators in the US have declared that ‘dumb’ pipes isn’t the way they want to end up, and they are getting more vocal about their plans to migrate to SDN and to make NFV (Network Functions Virtualisation) work in their networks. These aren’t niche upstarts by the way – we’re talking about AT&T, Verizon, EarthLink, Century Link and others.
One of the gaps that operators are trying to close is the innovation lag that prevents them from keeping up with the pace of disruptive new OTT services. Multi-vendor, multi-silo infrastructures result operator delays when launching new services of about 18/24 months.
OTT players, meanwhile, are at the stage where they don’t so much ‘launch’ services as conduct a perpetually evolving series of service experiments, listen to the user response and choose the one(s) they want to push harder.
Disaggregating the stack
The prevailing SDN strategy for traditional operators is disaggregation of vendor-driven technology stacks. This is not so much about breaking down the vertical integration of proprietary software and hardware elements, as enabling far greater flexibility over how this integration happens between best-of-breed components in an open network architecture.
It attacks the constraints of vendor lock-in, and as a resut more and more vendors are recognising that this has to be the way to go.
Take the components of a broadband network as an example, specifically the service provider OLT (Optical Line Terminal) for delivering FTTH. The chassis, the line cards, the software are all parts of a single solution. The networking functions they perform will inevitably become virtualised over time, but NFV today is still a long way from touching every part of operator infrastructure.
In many cases therefore, as soon as an operator decides to implement a given solution across its network they are at the behest of a single vendor’s roadmap to develop that technology further. They commit to embedding that technology into automated service delivery processes, B/OSS and the rest of it, all increasing the cost and risk of switching away from that vendor in future.
Compounding the inertia is the lack of commonality in the proprietary code found across multiple vendor environments, and the huge pressure this puts on internal development organisations to get automation to work in the first place.
Simple Network Management Protocol (SNMP)’s arcane complexity, non-human readable format and support for vendor-specific extensions is partly to blame here. The SDN-ready alternative lies in standardising on the YANG data model.
More flexible and efficient protocols such as NETCONF, coupled with YANG data models can be more open, and are far more aligned to common programming approaches, and being human-readable, aiding more rapid development cycles to level the playing field against OTT. Taking an open standardised approach to the main stay data models can go a long way towards elimination of programming duplication, and significantly shorten the distance to the hordes of agile OTT providers.
Opportunities to transition
Changing to SDN/NFV is challenging but there are opportunities for all operators to embrace the benefits as they evolve their services and take advantage of new copper, fibre and wireless technology. This opportunity is most pronounced in the developed broadband markets around the world. As we are bearing witness to the rise of “Gigabit Societies”, those operators who are enabling this through the supply of gigabit services, have a once in a decade opportunity to side step the status quo, and instead embrace the new paradigm.
The widespread move from vectored DSL to G.fast services is happening around the world as operators look to complement their long-term FTTH strategies with a pragmatic use of existing copper assets. The G.fast architecture was modelled around using SDN principles, which means that adopting the Broadband Forum WT-355 open YANG data model principles ensures easier management of multi-vendor components, and better integration with automated service delivery processes, than would ordinarily be the case.
In optical networks, we are witnessing the demand for a new generation of GPON technology to deliver the next leap in bandwidth from standard 2.5Gbps services towards more ubiquitous Gigabit services delivered on a 10G PON bearer.
Emerging standards, such as NGPON2 and XGS-PON stand to benefit greatly from SDN and NFV principles, permitting significant disaggregation of the hardware and software components, to dynamically build, horizontally scalable Ethernet based platforms, which can utilize the high availability of data center server clusters, to outperform traditional vertically integrated chassis approaches..
Finally, in the realm of wireless, the ratification of 5G services may still be a couple of years off but all the signs are that like with 4G, the NFV and SDN principles will be applied on even greater scales. As with the next generation copper and fibre access technologies that bring both the gigabit access capacity and the SDN and NFV foundations, the latest iteration of the 802.11ac Wi-Fi standard (Wave 2) can offer similar potential. Centralised proactive RF management, on top of a centralised SDN controlled, OpenFlow switching environment will prove key to ensuring Wi-Fi does not become the Gigabit Society bottleneck,.
Operators need to acknowledge these once in a decade opportunities and choose the path to smart or to dumb. ‘Dumb pipes’ is a legitimate business model, like power distribution and other utilities, but smart is where the margins are. Smart means a network built in software. Smart means SDN.
The author of this blog is Ronan Kelly, CTO of ADTRAN, EMEA & APAC and recently elected President of the FTTH Council Europe.
Comment on this article below or via Twitter: @ VanillaPlus OR @jcvplus