@fibresystemsmag | www.fibre-systems.com


FEATURE SDN


confidence in transport SDN while accelerating its development (see page 6). ‘Te goal is to leap to the end point by


providing the key parts of the future system for a carrier-style SDN-powered WAN, and thereby demonstrate conclusively the macro SDN service cases that people imagine can be delivered,’ says Janz. Te testbed will also help carriers determine how best to migrate their networks to SDN. Infonetics Research’s survey found that


inter-data centre and cloud services will be among the first transport SDN services to be deployed by the end of 2014. Te services will exploit SDN’s ability to


provision rapidly network resources when bandwidth is needed. For example, a customer could request a higher capacity link to enable a


The industry expects it will take time before transport SDN is deployed across the WAN


storage transfer, sending a large database between data centres. Running the same database from two centres and keeping it in synchronisation is another; here, link latency is key. Online retailers ordering extra cloud computing resources prior to a holiday season is another example. All the cases involve bandwidth requests set up


at short notice and used for the application duration only. Tey also promise operators new revenue opportunities. ‘It is all about using the WAN network as a programmable resource,’ says Elbers. ADVA Optical Networking working with IBM


has set up a multi-data centre SDN testbed using switching, server and storage technology from IBM and optical equipment from ADVA. Te testbed includes open-source SDN controllers from FloodLight and the OpenDaylight Project – an industry initiative developing an open- source SDN controller based on the Linux operating system – to demonstrate data centre and WAN orchestration in various use cases. Provisioning capacity as required to connect


resources could benefit carriers with room in their central offices. ‘You can put servers in the central office and essentially have a distributed data centre, where the compute is physically close to the consumers,’ says Intune’s Brandon. ‘SDN can also bring an awful lot of benefit to


Issue 1 • Autumn 2013 FIBRE SYSTEMS 21


TV broadcasting,’ says Dominic Elliott, UK CTO for service provider and media, Cisco Systems. Broadcast technology is still based on time


division multiplexing, while broadcasters would like to benefit from IP networks: creating links and tearing them down as needed. ‘Doing that using today’s technology is cumbersome,’ says Elliott. ‘SDN and the promise of programmable interfaces can create one instruction set that gets pushed down into the network, making doing that quite efficient in the IP world.’


Transport delay


Huawei has announced an SDN-based IP radio access network solution and has demonstrated various SDN prototypes. Te vendor says it has on-going SDN trials with operators from China to Europe. ‘We see some early deployment of SDN in WAN starting this year, and most use-cases are in the access network, the mobile core, and some data centre interconnection cases where centralised control is of great help,’ a Huawei spokesperson toldFibre Systems. But, despite its promise, the industry expects it


will take time before transport SDN is deployed across the WAN. Significant challenges must be overcome first. One important issue dictating transport SDN’s success is ensuring that the ecosystem remains open. Service providers will not buy all the vendors’ SDN controllers, and might even develop their own. It is important that each vendor’s controller will work with other controllers so that a service provider is not forced to buy a particular vendor’s solution. ‘Is it really open? Is it providing access to all the


functions that the standard defines? Is it forcing incremental middleware or soſtware applications that the service provider may not want or need?’ says Liou. ‘I do not know if this is achievable; it all depends how this ecosystem evolves.’ Another issue is making transport SDN


carrier-grade. One benefit of the traditional, distributed control plane is its robustness, with control split across the network elements. Transport SDN, with its central controller, must be able to sustain failures. Te core role of the controller must also raise security concerns. Te SDN controller is also required to have the


right degree of abstraction. Enabling access to too much networking detail – the capability to control every switch and knob in the system – would make the system overly complex. New data centre builds occur regularly, while


existing sites undergo comprehensive equipment upgrades. Service providers, in contrast, have hugely complex networks with a range of equipment – some decades old. ‘Most carriers upgrade their network


infrastructure once or twice a year,’ says Cisco’s Elliott. ‘To get to a point where you have stable code; open interfaces; application programmable interfaces well-defined; and processes currently using SNMP [simple network management protocol] and all the other things upgraded for SDN – that is quite a way out.’ Tis factor, he concluded, will probably be the biggest barrier of all. l


Roy Rubenstein is editor of Gazettabyte. www.gazettabyte.com


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36