Briefing: Dynamic optical networks
Part 2: Wavelength provisioning and network restoration
How are operators using reconfigurable optical add-drop multiplexers (ROADMs) in their networks? And just how often are their networks reconfigured? gazettabyte spoke to AT&T and Verizon Business.
Operators rarely make grand statements about new developments or talk in terms that could be mistaken for hyperbole.
“You create new paths; the network is never finished”
Glenn Wellbrock, Verizon Business
AT&T’s Jim King certainly does not. When questioned about the impact of new technologies, his answers are thoughtful and measured. Yet when it comes to developments at the photonic layer, and in particular next-generation reconfigurable optical add-drop multiplexers (ROADMs), his tone is unmistakable.
“We really are at the cusp of dramatic changes in the way transport is built and architected,” says King, executive director of new technology product development and engineering at AT&T Labs.
ROADMs are now deployed widely in operators’ networks.
AT&T’s ultra-long-haul network is all ROADM-based as are the operator’s various regional networks that bring traffic to its backbone network.
Verizon Business has over 2,000 ROADMs in its medium-haul metropolitan networks. “Everywhere we deploy FiOS [Verizon’s optical access broadband service] we put a ROADM node,” says Glenn Wellbrock, director of backbone network design at Verizon Business.
“FiOS requires a lot of bandwidth to a lot of central offices,” says Wellbrock. Whereas before, one or two OC-48 links may have been sufficient, now several 10 Gigabit-per-second (Gbps) links are needed, for redundancy and to meet broadcast video and video-on-demand requirements.
According to Infonetics Research, the optical networking equipment market has been growing at an annual compound rate of 8% since 2002 while ROADMs have grown at 46% annually between 2005 and 2009. Ovum, meanwhile, forecasts that the global ROADM market will reach US$7 billion in 2014.
While lacking a rigid definition, a ROADM refers to a telecom rack comprising optical switching blocks—wavelength-selective switches (WSSs) that connect lightpaths to fibres —optical amplifiers, optical channel monitors and control plane and management software. Some vendors also include optical transponders.
ROADMs benefit the operators’ networks by allowing wavelengths to remain in the optical domain, passing through intermediate locations without requiring the use of transponders and hence costly optical-electrical conversions. ROADMs also replace the previous arrangement of fixed optical add-drop multiplexers (OADMs), external optical patch panels and cabling.
Wellbrock estimates that with FiOS, ROADMs have halved costs. “Beforehand we used OADMs and SONET boxes,” he says. “Using ROADMs you can bypass any intermediate node; there is no SONET box and you save on back-to-back transponders.”
Verizon has deployed ROADMs from Tellabs, with its 7100 optical transport series, and Fujitsu with its 9500 packet optical networking platform. The first generation Verizon platform ROADMs are degree-4 with 100GHz dense wavelength division multiplexing (DWDM) channel spacings while the second generation platforms have degree-8 and 50GHz spacings. The degree of a WSS-enabled ROADM refers to the number of directions an optical lightpath can be routed.
Network adaptation
Wavelength provisioning and network restoration are the main two events requiring changes at the photonic layer.
Provisioning is used to deliver new bandwidth to a site or to accommodate changes in the network due to changing traffic patterns. Operators try and forecast demand in advance but inevitably lightpaths need to be moved to achieve more efficient network routing. “You create new paths; the network is never finished,” says Wellbrock.
“We want to move around those wavelengths just like we move around channels or customer VPN circuits in today’s world”
Jim King, AT&T Labs
In contrast, network restoration is all about resuming services after a transport fault occurs such as a site going offline after a fibre cut. Restoration differs from network protection that involves much faster service restoration – under 50 milliseconds – and is handled at the electrical layer.
If the fault can be fixed within a few hours and the operator’s service level agreement with a customer will not be breached, engineers are sent to fix the problem. If the fault is at a remote site and fixing it will take days, a restoration event is initiated to reroute the wavelength at the physical layer. But this is a highly manual process. A new wavelength and new direction need to be programmed and engineers are required at both route ends. As a result, established lightpaths are change infrequently, says Wellbrock.
At first sight this appears perplexing given the ‘R’ in ROADMs. Operators have also switched to using tunable transponders, another core component needed for dynamic optical networking.
But the issue is that when plugged into a ROADM, tunability is lost because the ROADM’s restricts the operating wavelength. The lightpath's direction is also fixed. “If you take a tunable transponder that can go anywhere and plug it into Port 2 facing west, say, that is the only place it can go at that [network] ingress point,” says Wellbrock.
When the wavelength passes through intermediate ROADM stages – and in the metro, for example, 10 to 20 ROADM stages can be encountered - the lightpath’s direction can at least be switched but the wavelength remains fixed. “At the intermediate points there is more flexibility, you can come in on the east and send it out west but you can’t change the wavelength; at the access point you can’t change either,” says Wellbrock.
“Should you not be able to move a wavelength deployed on one route onto another more efficiently? Heck, yes,” says King. “We want to move around those wavelengths just like we move around channels or customer VPN circuits in today’s world.”
Moving to a dynamic photonic layer is also a prerequisite for more advanced customer services. “If you want to do cloud computing but the infrastructure is [made up of] fixed ‘hard-wired’ connections, that is basically incompatible,” says King. “The Layer 1 cloud should be flexible and dynamic in order to enable a much richer set of customer applications.”]
To this aim operators are looking to next-generation WSS technology that will enable ROADMS to change a signal’s direction and wavelength. Known as colourless and directionless, these ROADMs will help enable automatic wavelength provisioning and automatic network restoration, circumventing manual servicing. To exploit such ROADMs, advances in control plane technology will be needed (to be discussed in Part 3) but the resulting capabilities will be significant.
“The ability to deploy an all-ROADM mesh network and remotely control it, to build what we need as we need it, and reconfigure it when needed, is a tremendously powerful vision,” says King.
When?
Verizon’s Wellbrock expects such next-generation ROADMs to be available by 2012. “That is when we will see next-generation long-haul systems,” he says, adding that the technology is available now but it is still to be integrated.
King is less willing to commit to a date and is cautious about some of the vendors’ claims. “People tell me 100Gbps is ready today,” he quipped.
Other sections of this briefing
Part 1: Still some way to go
Part 3: To efficiency and beyond