The changing optical transport market
by Michael Howard
The optical transport market is on the cusp of major changes, the beginnings of which are clearly in view today. Over the next few years, we’ll see emergence of a new optical transport layer-fused Ethernet-WDM packet transport with circuit-like capabilities. The services layer above the Ethernet-WDM transport will be simplified to IP/MPLS/Ethernet. This is the goal of most next-generation networks, also known as IP network transformation projects: Simplify the network layers, use fewer technologies, move to all packet. This means gradually reducing dependence on SONET/SDH as transport and ATM as part of the services layers. But first, the big picture.
It’s been more than five years since the great telecom bubble collapse, and service providers worldwide have their finances in order, pleasing investment centers by keeping their annual capital expenditures (capex) roughly at 15% to 17% of revenues. Since the world economies are perking along at sane if gradual growth levels, service provider revenues are rising gently each year, providing a lightly expanding capex pool for telecommunications equipment suppliers. Infonetics expects that worldwide service provider revenues will grow from US$1.4 trillion in 2006 to US$1.6 trillion in 2010, while capex grows from US$239 billion to US$265 billion in 2009, then drops back to US$260 billion in 2010.
But while revenue and capex will grow at a single-digit annual pace, traffic is growing strong at 50% to 100% per year, in no small part based on Internet video news stories and new video applications such as YouTube and social networking phenomena such as MySpace. Traffic is poised to increase even more dramatically as video and the personalization of video services come into the market over the next five to ten years.
Thus service providers face mounting challenges, mostly fuelled by the growing world dependence on broadband and mobile data for voice and video applications. Major Internet exchanges in Europe have seen daily traffic volumes increase by 70% per year (e.g., Amsterdam AIX, June 2006 to June 2007); Moscow’s volumes have grown 300%. By comparison, Japan’s growth has been 50%.
Competitive providers are assertive in rolling out new services in most world regions, causing incumbents to step up to deliver new services, add capacity to accommodate broadband traffic growth, and mount campaigns to attract a greater number of subscribers. As a result, broadband subscribers worldwide should grow from more than 230 million in 2006 to 450 million in 2010 (via a combination of DSL, cable, PON, and Ethernet FTTH). Subscriber bandwidths will increase from a few megabits per second to much higher speeds of 10, 30, 50, or 100 Mbit/s, depending on the world region and local providers. In the same period, mobile subscribers will grow from about 2.5 billion to more than 4 billion, and a significant portion of these will move from less than 100 kbit/s phone calls to data and video sessions at 10 to 100 Mbit/s.
The issue for service providers and challenge for equipment manufacturers and chip/component suppliers is that capex grows much more slowly than traffic. This traffic puts a strain on current optical transport networks based on TDM-the mainstay SONET/SDH technology is inefficient for packet traffic and relatively expensive and slow to deploy and provision.
So SONET/SDH must go-but it will be a 15- to 20-year disappearing act. WDM equipment sales have been growing nicely for the past few years, and as SONET/SDH spending starts to decline in 2007, Infonetics expects service provider spending on WDM to slightly surpass SONET/SDH equipment investments in 2010.
WDM is the only technology today that can solve the basic traffic growth problem; most of the traffic is riding in packets, which brings Ethernet into the equation. But Ethernet needs circuit-like capabilities to displace and replace SONET/SDH-this is the purpose of new Ethernet transport tunnel protocols, including Provider Backbone Transport (PBT), Provider Backbone Bridging-Traffic Engineering (PBB-TE), T-MPLS, and switched VLANs.
The Ethernet transport tunnels will eventually allow service providers to provision an Ethernet pathway, say from London to Rome, whereby the traffic can stay in the optical transport layer for the London-Rome transit, and not pop up and back to routers (hairpinning) at each city along the way. This keeps traffic off the routers, meaning that fewer router ports and routers will be needed.
These realities inspired manufacturers to create new optical products that Infonetics is calling packet optical network platforms (PONPs). PONPs combine configurable parts of Ethernet switching, SONET/SDH add/drop multiplexing/crossconnect, WDM transport, and ROADM. And they will have Ethernet transport tunnel technology embedded.
PONPs allow service providers to deploy single-node configurations from all-TDM to all-packet, and so serve as migration platforms to get from today’s TDM transport to the eventual fused Ethernet-WDM optical transport sans SONET/SDH. The earliest PONPs in the market were the ECI XDM family and later the Cisco ONS 15454 MSTP. Nearly all manufacturers have PONPs available now or will soon; examples include the Alcatel-Lucent 1850 TSS, Ciena 4200RS, Fujitsu Flashwave 9500, Meriton 7200 OSP, Nortel OME 6500, Siemens hiT 7080 and 7300, and Tellabs 6300 and 7100 Nano.
Optical transport networks are changing. Service providers are in IP network transformation projects with the goal of removing layers from their networks to arrive at an IP/MPLS/Ethernet services layer over fused Ethernet-WDM optical transport. Although it will take 15 to 20 years, ATM and SONET/SDH will gradually disappear. Ethernet-in the form of Gigabit Ethernet (GbE), 10GbE, and later 40GbE and 100GbE-is the “atomic unit” of transport traffic.
Within the longer-term migration, service providers will use PONPs to move from today’s TDM to all-packet infrastructures. Ethernet transport tunnels will be an essential ingredient of the services layer as well as the optical transport layer, allowing the displacement of SONET/SDH as well as enabling Carrier Ethernet switches to displace some routers.
There are other potential technologies just emerging, such as optical burst switching, a marriage of WDM and Ethernet, that show promise for a future optical transport, but that is another topic.
Michael Howard is cofounder and principal analyst, Optical, Routing & Metro Ethernet, at Infonetics Research Inc. (www.infonetics.com).