Above: Watch Cisco Postcards from SDN Roadshow, UK
By: Giovanni Fruscio, Marketing Manager, Service Providers, Cisco
It was a sunny day in London during the SDN Roadshow! We were meeting key customers in our City Offices and we saw many people joining and starting good conversations already at the registration desk.
Our objective was to share our ideas with customers and partners on the benefits of Software Defined Network (SDN) architecture for the evolution of the network infrastructure, in terms of applications development on the network, service velocity and cost reduction. Further, as SDN is defined mainly as an architecture with control and data planes decoupled, the speakers have shown the additional benefits of the Cisco ONE (Open Network Environment), including Openstack, Openflow, Overlay Networks (and some other acronym, still starting with "O"). These are not mandatory and applicable or not to SDN architecture, but very helpful to manage the interconnection with Data Centers and Cloud heterogeneous and multi-vendor infrastructures.
But the characteristic that is mostly identified as the benefit of SDN architecture introduction is the network abstraction and the introduction of onePK API, being developed on most of the IOS, NX-OS and IOS-XR platforms and allowing developers, operators, engineers to program the networking devices. This is to explain service velocity (open API and developers community will decrease the time to develop and market new applications), but what about the Total Cost of Ownership (TCO) reduction?
The speakers went through a detailed explanation on distribution or centralisation of the control functions, the benefits of flexible allocation of resources versus the resources reservation needed by former architectures (especially for voice and real-time services) and, last but not least, Network Function Virtualization (NFV), where and when it makes sense. In fact, virtualization on its own will not reduce opex, unless a robust and efficient orchestration is developed (otherwise it's simply another layer to manage for all the systems) and the x86 architecture will be more efficient than ASICS for the specific need. As far as it concerns Data Center, the two main topics been developed were VXLAN (to interconnect L2 segments across L3 networks) and VEM (the virtual switch).
The final moderated Q&A session was surprising and we saw the majority of the attendees join the conversation, that lasted a record two-hours. The day after we had a workshop with a global customer and took advantage of a multipoint Cisco Telepresence bridge across 7 locations, discussing about the benefits of SDN, with a stronger focus on the protocol architectures being used and the evolution of the onePK interface, in terms of alignment with functionality available through CLI and roadmap.
Beyond the contents, the Telepresence experience was excellent as it allowed us to organise a meeting across seven different countries in just one week thus having an update in near-real time about Cisco's developments on segment routing and other emerging protocols and architectures. One of the other good things of Telepresence is that in most cases it's a shared resource and so the bridge expires at a certain time and forces everybody to stick to the agenda, timing as well as closing time.
This allowed everybody to come back to their daily jobs thinking about how to take advantage and make fruitful these new standards for telco and cloud infrastructures and myself back home flying across the rainy London afternoon …
Find out more about Cisco at SP Events (blogs, videos, photos) here.