November 2001

 
Focus On Flexibility: Meeting The Needs Of Emerging Network Applications 

BY TOM GREELEY

Packet transport technologies, including Internet Protocol (IP), frame relay, and asynchronous transfer mode (ATM), are quickly becoming the method of choice for delivering both control and bearer traffic on existing and emerging networks. A number of evolving network applications, including 3G wireless, SS7 gateways, and media gateways, are highly reliant on packet transport. This technology has established a stronghold in all network domains, from access to core. Emerging from the enterprise market, IP is now being adopted by public network providers. Packet transport technology brings with it the flexibility and scalability to handle demand surges within the network as new services are introduced. The circuit-switched technology found in legacy networks is much more restricted to fixed resources in that higher demand requires more line cards, power, and rack space.

ECONOMIC PRESSURE MOUNTS
Just as the enterprise market has influenced the public switched (or carrier) market, the reverse is also true. Carrier grade equipment and software delivering high availability for telecom voice and data services are finding their way into the enterprise market. Services such as VoIP are extending beyond being novel services aimed at saving money at the expense of quality. Enterprise customers are now demanding the same level of service and availability found in public switched networks. The dilemma faced by enterprise network and service providers is one of cost as carrier grade equipment and software is more expensive than enterprise-grade equipment. For the emerging service and network provider, it is very difficult to compete with existing public carriers that guarantee a high level of service. In today's environment, the least amount of buck for your high-availability bandwidth is the goal of every provider. In addition, the provider needs the ability to help deliver any new high-revenue-generating application or service as quickly as possible. Time to market is critical.

Along with current economic pressures and uncertainty as to the rollout of network services, telecom equipment suppliers no longer have the luxury of developing application-specific platforms. There is an immediate need in the telecom market for a multiservice platform that incorporates new open standards for packet transport and systems management, new packet transport and network processing technology, and storage-based solutions. In addition, as the edge of the network continues to move outward toward the enterprise space, high availability and carrier grade solutions will still be necessary elements of any new architecture. Scalability of network equipment will be essential in allowing providers to have a lengthy return on investment without the fear of major forklift upgrades. Next-generation telecom platforms will be expected to provide seamless integration of new technology and increased shelf capacity.

STANDARDS AND THE MULTISERVICE PLATFORM
Standards benefit the telecom industry by allowing multiple vendors to interoperate within a common framework. Standards also provide a product evolution roadmap for telecom equipment manufacturers. How quickly a manufacturer adopts a particular standard depends on the market's willingness to advance. There is always a transition gap from technology and equipment based on legacy standards to newly adopted, standards-based technology. It is usually more a question of when rather than should a standard be adopted. Timing the deployment of a new standard is based on risks related to the current state of maturity for a particular standard and the need for it in the market. Equipment manufacturers take the role of either a market innovator or follower, with risks accompanying either approach. A successful multiservice platform strikes a balance between established and well-accepted standards and technology innovation.

For systems management, Intelligent Platform Management Interface (IPMI) is gradually displacing architectures dependent on CompactPCI bus system management. PCI Industrial Computers Manufacturers Group (PICMG) has adopted IPMI for use in CompactPCI shelves, or chassis, as the PICMG 2.9 "System Management Specification" standard. Many telecom equipment suppliers have integrated support for the PICMG 2.9 standard in both CompactPCI chassis and boards. Customers are looking towards industry standards for chassis management because they provide multiple sources for both hardware and management capabilities. IPMI provides additional management capabilities such as more complete device capability reporting and diagnostics capabilities.

In many cases, standards emerge from the need for consensus within a marketplace. The need for packet transport technology, specifically IP, in telecom network solutions has grown exponentially. PICMG 2.16 "Packet Backplane Specification" has become the accepted standard for IP backplane transport on CompactPCI platforms. Solutions based on PICMG 2.16 are designed to be capable of supporting Gigabit Ethernet (GigE) channels in a highly available dual star configuration. The support of GigE will depend on the introduction of PICMG 2.16 compliant GigE fabric switches. Manufacturers of CompactPCI boards can deliver IP-based application blades that can be supported by any PICMG 2.16 platform. The architecture is ideal for control plane IP applications such as SS7 gateways. In addition, certain functions normally supported by the CompactPCI bus are capable of being implemented using PICMG 2.16, including diagnostics and boot along with download support. IPMI and PICMG 2.16 are able to help provide the systems management backbone for next-generation solutions.

A HIGH-SPEED DATA FABRIC
PICMG 2.16 is an integral piece of any next-generation platform but it is limited by protocol (IP only) and capacity (GigE over dual star). As applications approach the bearer traffic capacities found in the edge and core network domains, a true multiservice platform provides the scalability required to support these applications over a "protocol-less" high-speed data fabric. The "protocol-less" nature of this platform should be capable of supporting any packet-based technology including IP, frame relay, and ATM. By allowing point-to-point, or full mesh, high-speed interconnect across the backplane, native packet transport solutions can offer improvements of 100 to 1,000 times channel capacity over H.110 (the H.110 bus has a maximum capacity in the 250Mbps range, while the full mesh IP 100BaseT is in the 32 Gbps range with scalability to almost 320 Gbps with 1000BaseT). The integration of digital signaling processor (DSP) and network processor technology help provide the engines to drive applications over this high-speed data fabric.

An example application benefiting from a high-speed fabric would be a media gateway. A typical media gateway application will serve as an interface for a higher bandwidth packet network to a number of lower bandwidth TDM interfaces. They may be used in softswitch applications where packet switching replaces TDM-based switching. DSP boards perform the necessary processing to deal with the TDM world, including echo cancellation, compression, and packet conversion. In the case of modem or fax support, DSPs provide tone recognition and demodulation. The DSPs interface to a packet network. For VoIP applications, the PICMG 2.16 IP network can be used to carry the data traffic. A network processor provides the LAN to WAN interface, as well as traffic aggregation. For voice over ATM, the DSPs can interface directly over the high-speed data fabric. A multiservice platform that supports both PICMG 2.16 and a high-speed data fabric is ideal for addressing different types of media gateway applications.

STORAGE AND THE MULTISERVICE PLATFORM
A multiservice platform should also be capable of supporting storage-based applications requiring reliable high-speed connectivity to internal and external storage mediums. A means of providing this capability is through a fibre channel fabric. Fibre channel is a standards-based solution that offers higher reliability and throughput performance over SCSI for networking and data distribution. Manufacturers of computer equipment and storage devices have adopted fibre channel as a leading technology. With a fibre channel fabric, disc storage and media devices can be connected through fibre channel hubs or switches. In addition, the fibre channel fabric can connect CPUs with external HA-storage systems.

Through a multi-shelf system containing general-purpose CPU boards and disk storage, a multiservice platform can provide the development framework for an entire class of network applications including application servers, feature servers, call servers, and home location registers (HLR). The CPUs use Ethernet and the fibre channel fabric to share data. If a WAN interface is needed, network processor boards can be used for the I/O and translation into GigE.

Meeting the Needs of Emerging Network Applications
Five-nines or better availability is an essential attribute of most telecom network solutions. Components must adhere to stringent MTBF (mean time between failure) and MTTR (mean time to repair) requirements, while the supporting architecture must provide the software infrastructure to manage all components within a distributed high-availability model. A multiservice platform should not only be capable of supporting technology such as fibre channel, PICMG 2.16, and high-speed data fabric, but also provide the ability to monitor and manage all services effecting components. Redundancy between components assures that a system will meet five-nines. High-availability management software for a multiservice platform must easily integrate new components and technologies.

New opportunities across all network domains and technologies have driven the need for a single highly available platform solution. New standards, such as PICMG 2.9 and PICMG 2.16, are creating a consensus among vendors for system management and packet transport methods. Platform scalability has become a critical requirement as data throughput and processing needs continue to increase. This scalability is achieved through a high-speed data fabric with DSP and network processor technology. Server-based applications requiring storage have put an added emphasis on the option to deliver highly available integrated storage solutions on the same platform. The incorporation of all of these attributes into a highly available architecture results in a very flexible system framework. Depending on the boards and other system components deployed within this framework, a wide variety of existing and emerging network applications can be supported by a multiservice platform. Many uncertainties and risks associated with today's telecom market can now be overcome while setting the stage for what will be a pivotal moment in telecom history.

Tom Greeley is senior product manager of High Availability System Solutions for Motorola Computer Group's Telecommunications Business Unit in Tempe, AZ. Motorola Computer Group is a business unit of the Motorola Integrated Electronic Systems Sector (IESS). A leading supplier of embedded computing platforms for equipment manufacturers in telecommunications, network storage, imaging, medical equipment and semiconductor production and test equipment industries, the company can be found on the Web at www.motorola.com/computer.

[ Return To The November 2001 Table Of Contents ]