IP service velocity in next generation of optical metro networks
Service providers must choose to rely on existing network designs or offer Internet Protocol-optimized metropolitan-area infrastructures.
RUSS HERTZBERG, Zuma Networks
The race is on among carriers and service providers to build next-generation optical metropolitan-area networks (MANs). Separation of winners from survivors and losers will be determined by the business success or failure of Internet Protocol (IP) service offerings. Industry analysts are projecting that industry-wide carrier and service-provider data-services revenue will exceed voice-services revenue within the next two or three years. By 2002, burgeoning IP data traffic will account for 90% or more of the traffic load on worldwide telecommunications networks.
Network equipment vendors have teamed up with a group of greenfield service providers-no legacy time-division multiplexing (TDM) networks-to promote Gigabit Ethernet and 10-Gigabit Ethernet as the logical Layer 2 standard for MANs. Competitive interaction between these new service providers and the incumbents will be particularly interesting in the next few years. Incumbents must decide between relying on existing MAN designs or offering newer, IP-optimized MAN infrastructures.
Carriers and service providers are now focusing competitive differentiation efforts on expansion and improvement in MAN optical transports and related IP service offerings. Factors driving this focus include:
- IP traffic bottlenecks. Most urban and suburban business connections to the Internet are relatively low-speed TDM circuits. Residential connections are predominantly low-speed public-switched telephone networks with a gradual growth in DSL and cable modem. These so-called "residential broadband" services appear to be lagging behind actual consumer requirements and demands. Service coverage areas, installation timeliness, and delivered bandwidth all remain less than satisfactory to customers and prospects.
- IP traffic origin and destination concentration in dense-population areas. While difficult to measure, IP origin and destination traffic closely matches the density profiles of Internet and Web users and IP traffic-generating servers. End users and Web servers are heavily concentrated in major metropolitan centers.
- Concentration of large IP data-services customers and prospects in metro areas. Internet service providers, next-generation network providers, Web hosting/collocation providers, and major enterprises-potential key customers for MAN IP services-are ordinarily found within a metro area around a dense-population core.
- Central-office (CO) fatigue in metro areas. Incumbent local-exchange carrier (ILEC) COs support several generations of switching systems as well as new high-speed access systems like DSL access multipliers. Lack of physical room for new optical and IP switching systems, along with rigid equipment qualification requirements, make some COs less than ideal candidates for high-speed, optically enabled, IP-centric network points of presence.
- Available fiber plant in metro areas. Return on investment for network system upgrades is best in large cities with the highest possible customer density. Major fiber installation programs are ongoing in many cities, and installed, yet unused, fiber is readily available in many areas.
What are IP services? Simply put, they are billable services for consumers and businesses that service providers bring to market quickly for IP networks. Figure 1 shows the relationship between IP services and other network functions.
What are some of the IP services that can or should be offered on a next-generation MAN? The following list is a sample of an IP services menu for a focused MAN provider:
- Network access
- Virtual private networking
- Firewall and secure network services
- Traffic management and quality-of-service (QoS)
- Content delivery
- IP data-center services
- Web hosting
- Application hosting
- Managed network operations
What is a high IP service velocity MAN? Several criteria or conditions can be identified that enable such a MAN network:
- Highly scalable optical transport and switching systems.
- Rapid provisioning of new optical paths for IP traffic and services.
- QoS mechanisms within key network elements to meet the variable delivery sensitivities of different traffic types (packet data, voice, and video).
- Scalable wire-speed switching and routing systems.
- Scalable IP service delivery platforms supporting rapid service introduction and ongoing modification.
- IP service provisioning systems that support network operator-controlled, automatic, or subscriber-controlled modes of operation.
These criteria suggest that the ILECs carve out a portion of their SONET/ATM infrastructure for redeployment as the next-generation MAN optimized for IP. Given the trends in traffic types and overall service-provider revenue, to do otherwise is to run the risk of providing permanent competitive cost and service advantages to the newer service providers.
Figure 2 shows the overlay of a new optical IP ring onto an existing metropolitan or regional SONET implementation, with equipment configurations optimized to meet the suggested criteria for a high IP service velocity MAN. The multiplexing of wavelengths over optical fiber (WDM) has expanded the bandwidth capacity of existing and new fiber exponentially. New optical transport and switching systems permit the instant addition of optical paths, one wavelength at a time, as traffic needs dictate. Additional developments in optical transport and switching promise to increase the capacity and flexibility of fiber even more. For example, leading metro optical-switch providers can now offer multiprotocol wavelength-sharing services, permitting service providers to run multiple services from multiple subscribers over a single wavelength.
The pervasiveness of IP connectivity means that new services need to be able to offer enough bandwidth to provide for a satisfactory user experience in the event of unpredictable runaway demand. Hence, service providers need the tools to provision new wavelengths for one or more subscribers with one-day notice or less. It also means service providers should have the flexibility to allocate a pool of wavelengths for services such as Web casting, with multiple short-term subscribers moving in and out of the pool on demand.
The QoS needs for a MAN implementation is determined by the mix of services offered. In the case of a combination of packet data, voice, and video, the core switch/routing platforms need to police and prioritize the micro-flows that contain different Layer 4 and higher protocols-at wire speed. Gigabit Ethernet and 10-Gigabit Ethernet ports on these systems will be supporting hundreds or thousands of logical flows, so a traffic classification system that operates at a port level of granularity will not be able to identify and isolate the flows that require priority treatment.
Wire-speed routing on every Ethernet port in a modular (expandable) Layer 2/3/4 system ensures subscribers get the IP routing performance and scalability needed to efficiently handle sudden surges in IP traffic. Matching of performance and scalability at Layer 2 and Layer 3 permits balanced network expansion.
Scaling the Layer 2/3 switching function creates demand and momentum for scaling of higher-level IP services. Without a platform for scaling these IP services, MAN providers may find themselves (or their suppliers) operating expensive, space-consuming, general-purpose server farms with separate systems for clustering and load balancing. New platforms are coming to market that permit the matched scaling of IP services with Layer 2/3/4 switching and routing (see Figure 3).
Integrated Layer 2/3/4 switching/IP service platforms can aggregate the Layer 2/3/4 switching and routing with a wide variety of IP services in an open system. IP services can be clustered and expanded within the chassis-programmatically tied to the services of the Layer 2/3/4 fabric-or integrated with existing or new operational support systems using general-purpose development tools.
Bandwidth and circuit provisioning in TDM networks now appears unreasonably slow in light of the rate of traffic growth driven by the "Internet phenomenon." Next-generation MAN providers have delivered provisioning systems at the opposite end of the spectrum-permitting instant customer control of the bandwidth faucet.
Provisioning for next-generation MANs needs to vary by service and organizational policy. A service provider selling a standard entry-level Web hosting service should be able to automatically provision bandwidth and network resources. A collection of technology companies holding random Webcasts needs to self-provision bandwidth on demand. A larger organization purchasing managed network services may wish to carefully control cost increases resulting from growing bandwidth or service needs.
With IP traffic continuing to double every 100-200 days, the providers of next-generation MANs can expect an ongoing challenge meeting unpredictable, unconstrained network-capacity growth requirements. The bursting of the "dot-com bubble" and a return to Internet technology realism does not change fundamental trends in telecommunications network traffic. Service providers who prepare and execute MAN build-outs that can sustain rapid introductions of new IP services will be the short- and long-term winners. Radical scalability along each critical dimension of MAN network services is the best way for MAN service providers to position themselves for double-digit revenue growth opportunities in IP services.
Russ Hertzberg is vice president of marketing at Zuma Networks (West Hills, CA). He can be reached via the company's Website, www.zumanetworks.com.