AT THE FOUNDATION of all long distance data communications networks we find optical fibre. Fibre has the multi-terabit capacity, and multi-thousand kilometre reach, needed to satisfy internet demand on a global basis and at the right cost points for service providers to maintain profitability. In order to light that fibre we rely on a technology called Dense Wavelength Division Multiplexing (DWDM) to provide the raw, point to point capacity. That capacity has to be managed, protected and kept at peak efficiency, so that’s typically where an electronic switching technology comes in. OTN electronics allow service providers to keep track of the status of services, to provide sub-50ms protection or shared restoration and to electronically gather together lower data rate services so that wavelengths are filled efficiently. Together they make up “the transport network”.
In a recent report on the trends for transport network architectures (TNA), Andrew Schmitt, of the analyst firm Infonetics, surveyed service providers to understand the way that existing and emerging TNAs are gaining traction.
The first thing to understand is what are the options in terms of TNAs and the business objectives for service providers. A “traditional” transport network consists of DWDM transmission boxes and OTN switching boxes. IP routers plug into the OTN layer and see digital capacity as a result. This architecture is widely used, but the use of two separate platforms for transmission and switching is inefficient both financially and operationally. So the modern version of this “IP over OTN” architecture involves the integration of non-blocking OTN switching into a high capacity DWDM transmission platform, and connecting routers to that platform using low cost, short reach optics. An alternative architecture brings the DWDM coloured wavelengths into the router itself simplifying the architecture, and puts the IP engineers back in control of their capacity needs. The routers provide all of the functions that would otherwise have been handled by the OTN switch.
The marketing battle wages over which architecture is best, but what are the business objectives for the service provider in deploying these architectures?
Minimise the total cost of ownership for
the entire network
Improve the response time for new
capacity and services to be available
Be aware that there is an existing
investment in transport network
technology, and equipment in this part
of the network is typically depreciated
over eight to ten years, compared with
router equipment that is typically
depreciated over three to five years.
Additionally, the report suggests that there are four key requirements in a transport network architecture; Agility, Flexibility, Simplicity and Resiliency.
The key value of the Infonetics report is that it reveals what service providers are actually doing in their transport networks:
Networks get “meshier”
While Sonet networks were historically built as resilient rings, modern transport networks are designed as a mesh – either full or (more commonly) partial. Mesh networks can offer the same resilience as rings, but provide diverse routing for more agile operational flexibility. Critics would say that a meshed all-optical network is more complex operationally, but by integrating OTN switching with optical transmission this drawback is eliminated. As a result it’s no surprise that 95% of the service providers surveyed say they have moved, or will move to meshed transport networks.
Transport platforms will converge functions
One of the biggest advances in recent years is the merging of the DWDM transmission and OTN switching platforms, eliminating boxes from the network and avoiding the need for every service to be manually patched into a separate OTN switch. The result is lower capex and opex, and 90% of those surveyed say they have deployed, or plan to deploy, an integrated OTN/DWDM platform.
OTN protection gets smarter and more efficient
Sonet 1+1 protection effectively defined the gold standard for network resilience, but at the price of requiring that every Gb/s of working service capacity would need an equal amount of protection capacity that would just sit there “waiting” for a failure to occur. Innovations such as Fast Shared Mesh Protection (FastSMP™) now mean that sub-50ms protection can be provided by the OTN layer using shared, not dedicated capacity. In the survey, 70% of service providers indicated that they plan to use OTN Shared Mesh Protection.
ROADMs will become part of a multi-layer switching architecture
Historically Reconfigurable Optical Add Drop Mux (ROADM) devices provided useful flexibility in the network, but were limited by contention within the ROADM architecture while the so-called CDC (Colourless, Directionless, Contentionless) ROADM is still an aspiration. Integrated OTN switching provided a “digital ROADM” capability that continues to be very popular amongst service providers. The advent of coherent super-channels at 500Gb/s today, and terabit scale in the future, may give a new lease on life to the ROADM in order to create super-channel “express lanes” in very high capacity backbone networks. By combining the super-channel ROADM with integrated, non-blocking OTN switching service providers can drive down the capital cost of transport networks without a negative impact on network flexibility.
Software defined networks “go carrier grade”
Control planes have always been accepted as a great way to lower opex in service provider networks, and GMPLS has proved itself as a carrier-grade, distributed control plane. The Software Defined Network (SDN) architecture is now making its way from the data centre, and into the transport network in the form of Carrier SDN (being developed by the Open Networking Foundation).
These companies must continue to adopt new vendors, equipment, and network architectures, while adding the features and services customers demand. Technologies like metro ROADM in the 2000s, and coherent WDM today are good examples; IPoDWDM is not.
Instead the favoured architecture will make use of widespread multi-layer networking built on an intelligent optical transport mesh. This will deliver low cost, highly managed digital servers to the more expensive edge router layer, whose job it is to perform high value packet processing and to deliver the required service intelligence at the edge of the network.