When the telecommunication industry recovers from the current economic climate, a different market landscape will emerge, and new test and measurement requirements will be needed.
April 1, 2002
I n the middle of what is probably the worst recession the telecommunications business has experienced in its 100-year history, two questions come to mind: When will the market recover a sustainable growth profile, albeit lower than the boom years of 1999 and 2000? And, how much will the market landscape have changed in terms of the players and the preferred technologies and architectures when recovery is established?
The general answer from analysts to the first question is early 2003, particularly in metro networks and edge IP router markets, though long-haul transmission may take a little longer due to excess bandwidth in the network.
The question of preferred technologies and architectures — and the new test and measurement requirements that will develop as a result — requires a more detailed analysis.
EVOLUTION OF METRO AND CORE NETWORKS
Although investment in capital equipment was cut during 2001, particularly in North America, and will decline further on a global basis in 2002, telecom traffic continues to grow. Market research firm, RHK, estimates that Internet traffic continues to double annually and will gradually absorb the excess network capacity installed in 2000. Goldman Sachs, McKinsey & Co. recently analyzed the North American transport market1 and estimated that 22 per cent of installed long-haul fiber is currently lit, and between 20 and 30 per cent of available lit capacity is being used. This led the analysts to estimate a one- to two-year delay before long-haul investment restarts.
The metro and edge equipment market is expected to recover much more quickly, probably in the second half of 2002. This is expected because the edge network remains a bandwidth bottleneck to the steadily increasing level of data traffic, particularly as wideband access continues to grow and Gigabit Ethernet (GbE) emerges as a favoured access technology for enterprise networks.
Today, the capabilities of the core and metro/edge networks are highly mismatched. One estimate puts the aggregate capacity for metro/edge networks in the U.S. at 25 to 30 Tb/s (including voice access), while the long-haul core network has a capacity of several hundred Tb/s. While these figures may not be strictly comparable, the metro network as presently constructed does not allow full use of the potential offered by the core network. In the future, more traffic will stay in metro areas (possibly as much as 70 to 80 per cent) due to local content servers and caching of Web sites.
In 2002, many service providers will focus on survival and consolidation in view of the enormous burden of debt coupled with price/tariff erosion. Cost reduction is the order of the day, so new deployment of network equipment will have to reduce unit costs and open up new revenue opportunities. Some key objectives include: reducing the cost of transmitting and switching a bit by a factor of ten; looking for more integrated equipment that reduces footprint, power consumption and interconnection costs; removing or integrating protocol layers to reduce equipment and management costs of a multi-layer network; and developing more effective ways of guaranteeing Quality of Service (QoS) in data networks to offer added value to customers.
Core networks will mostly focus on reducing the cost per bit by greater use of optical transmission and switching, and eliminating electronics. More flexible and faster provisioning schemes using mesh networks will probably emerge. Operators of core networks may choose to be either retailers of high-bandwidth pipes (and wavelengths) or wholesalers to ISPs and metro service providers.
Metro networks must expand to accommodate bandwidth demand for storage area networks and content storage, as well as provide Gigabit Ethernet access. Emphasis will again be on low unit costs, which will take full advantage of the specific topography and shorter distances of the metro network.
Figure 1 (page 25) shows some of the likely technologies appearing in metro and long-haul networks over the next four to five years. In metro networks, native Ethernet services are expected to see high growth, particularly Gb and 10 GbE. To reduce unit costs, some operators plan optical Ethernet infrastructure with Ethernet switches directly connected over fiber, avoiding additional encapsulation with SONET/SDH and ATM.
In core networks, the transport and switching will migrate more towards optics. There are already signs of this with several manufacturers offering ultra long haul DWDM systems that eliminate electronic regenerators for distances of several thousand kilometres. As bandwidth demand increases, the granularity of bandwidth provisioning will increase. Today, the smallest increment of bandwidth in the core is typically STM-1. Over time, this will rise to STM-16 and eventually STM-64, which are the typical line-card rates for a core IP router. Cross-connecting these channels would be achieved with a superband DCS or an Optical Cross-connect Switch (OCS). The latter could either have an electrical core (OEO), or a non-regenerative pure optical switch (OOO). The OOO switch would only deal with wavelength granularity.
Figure 2 shows the likely architecture of networks by 2005:
The core network consists of transparent optical subnets, which are managed at the wavelength level, typically carrying 10 Gb/s or 40 Gb/s payloads. The network is mesh rather than ring, with protection/restoration provided by dynamic re-routing under direction from an IP control plane.
The current view is that this bandwidth provisioning and restoration would use GMPLS (Generic Multi-Protocol Label Switching), extending the routing protocols of the IP world into the optical transport and switching network. Switching and transport will become more integrated — for example, routers and Ethernet switches may also incorporate DWDM transceivers for direct connection to the fiber network. Switching can be implemented in circuit-switch technologies such as DCS or OCS, or can be implemented in the packet domain through label-switched IP routers or ATM and Ethernet layer 2 switches. Label-switched paths (LSPs) or end-to-end tunnels can be created using the GMPLS control plane.
Architecturally, any of these network elements can be represented by a data or forwarding plane that handles the traffic, and an IP-centric control plane that implements the routing protocols. (Please see Figure 3).
In general, networks can be divided into the control/management plane and the transport/switching data plane. The data plane can be further sub-divided into optical and electrical functions.
TEST AND MONITORING REQUIREMENTS
Over the coming years, the telecom network will “de-layer” to some extent, reducing the levels of encapsulation in the protocol stack and, correspondingly, the diversity of equipment and interconnections that add to capital and operating costs. Network equipment will embrace a wider range of functionality, and with the emergence of Ethernet and MPLS, some functions that were implemented in hardware will instead be activated through Layer 2 and Layer 3 software.
What does this mean for the test equipment and monitoring systems used to install and maintain these networks? In common with de-layering of the network, there will probably be some consolidation of test capability and merging of functions and protocol layers of test. What is in a test set will also be determined by which team is using it. In terms of the network layers, it is possible to segment the test capabilities that will be required and what emerging technologies will drive those capabilities. (Please see Figure 4).
Although many of the same technologies will appear in both metro and core networks, the application requirements to be tested will differ. In general, edge/metro networks provide a diversity of interfaces between Provider Edge (PE) and Customer Edge (CE) offering managed services and Service Level Agreements (SLAs).
The metro/edge network also aggregates traffic into higher capacity flows, which can be allocated to a specific TDM pipe or wav
elength (Layer 1 provisioning) or to a GMPLS tunnel also called a Label Switched Path (LSP or Layer 2 provisioning). The core network only deals with aggregated flows and relatively few classes of service. This enables it to expedite forwarding and minimize cost per bit. This differentiation will lead to different test requirements in the coming years. (Please see Figure 5).
The key disruptive technologies likely to change the telecom landscape in the next few years are native Ethernet, transparent optical networking and an integrated IP-based control plane. There will probably be less electrical regeneration and fewer protocol layers (less encapsulation). While a lot of attention will focus on these new technologies, many service providers must maintain and enhance their legacy networks — as these continue to be the main source of revenues. Test equipment will have to accommodate the older networks as well as the new disruptive technologies.
As the network functions merge between switching and transmission, as well as protocol layers, test equipment will provide new combinations of functionality to reflect the changing network architecture and equipment capabilities.
Hugh Walker has been with Hewlett-Packard Ltd. and Agilent Technologies for over 30 years, involved in R&D and marketing of telecommunications products. He is currently market research manager at Agilent’s telecommunications division in Scotland.
1 U.S. Communications Infrastructure at a Crossroads: Opportunities Amid the Gloom, August 2001
|Parametric testing of installed fiber (dispersion, loss)|
|High density WDM resolution.|
|Impairments in transparent optical subnets (DWDM and Optical Cross-connect switches)|
|Move to more transparent optical networking and ultra long haul transmission reduces opportunity for electrical test and places greater emphasis on optical parametric measurements|
|(transport and switching)|
|Higher bit rates (2.5/10 to 40/80 Gb/s)|
|Framing from SONET/SDH to POS to OCh (G.709 digital wrapper)|
|Virtual concatenation measurements, parallel measurements in next generation SONET/SDH bandwidth managers|
|GbE, 10 GbE and RPR measurements to Layer 2|
|Check activation of SLAs and protection/restoration algorithms|
|Lower cost per bit drives evolution to higher bit rates per channel and router line card|
|Digital wrapper forward error correction assists economics of optical subnets|
|Growth of data requires data-enabled (Ethernet) edge equipment for improved aggregation|
|Ethernet moves from edge into core, as native Ethernet services expand|
|Managed Ethernet services at Layer 2 (VLAN tag/MPLS)|
|Resilient Packet Ring (RPR), Ethernet and MPLS move provisioning and protection from Layer 1 to Layer 2|
|Unified control plane (GMPLS) for routers and switches including edge devices|
|Check inter-working (protocol test) between routers and also routers to OXCs and Ethernet switches|
|Confirm SLAs, QOS priorities|
|Need to reduce number of separate control planes to reduce Opex. Desire for multi-vendor multi-technology networks with integrated control plane|
|Dynamic bandwidth provisioning and rapid mesh restoration using GMPLS|
|METRO TEST NEEDS||CORE TEST NEEDS|
|Installation and turn-up of service interfaces between CE and PE, mainly 10/100/1000E|
|Check of traffic mapping and QoS attributes between service interfaces (SLA) and aggregated flows to core network (2.5/10 Gb/s)|
|Installation and maintenance of pure-play Ethernet networks relying on Layer 2 for provisioning and restoration (1/10 GbE, RPR)|
|Checking performance of managed Ethernet services using VLAN tag (802.1p/q)|
|Testing next-generation SONET/SDH aggregation multiplexers and associated metro DWDM transmission|
|Qualifying performance of metro wavelength services|
|Optical parametric measurements of installed fiber to check support of 10 and 40 Gb/s long-haul transmission and transparent optical networks|
|SONET/SDH, Packet over SONET, Digital Wrapper (G.709) and 10 GbE framing for transport and switching|
|GMPLS control plane inter-working tests|
|Installation and turn-up of wavelength and electrical cross-connects down to granularity of STS-1/STM-1|