July 1999

Next-Generation Campus Networking

BY TONY RYBCZYNSKI

Evolving towards a single, unified infrastructure for telephony, data, and video — that’s the challenge for managers of campus networks. A challenge that could scarcely be more urgent, for fulfilling it is essential if the enterprise is to improve customer service, support process reengineering, and address globalization issues. It is, moreover, a challenge that the enterprise will have to meet while managing operational and lifecycle costs, even as networking in the campus environment admits greater complexity.

THE CURRENT GENERATION
To appreciate the extent to which their networks must evolve, enterprises should consider the limitations of the current installed base. This base, as enterprises are already learning, constitutes a bottleneck, in terms of reliability and performance, a bottleneck that will prove ever more constricting to enterprises as they approach business-critical applications such as IP telephony.

With an application such as IP telephony, the shortcomings of the installed base become apparent. These campus network infrastructures may be seen as being incompatible with IP telephony, since they lack the ability to differentiate between voice and data applications (at least, not without significantly impairing performance), and since they lack the requisite reliability (barring costly equipment duplication).

The limitations of the installed base affect campus networks of all sizes, from the small (a couple of workgroups) to the large (hundreds of workgroups, handled by multiple campus switches), from a single building (with, perhaps, a single switch) to a multi-building environment (with multiple switches working into a WAN edge device, a router, say, or an enterprise network switch).

The limitations of the installed base will become apparent to all sorts of companies that have campus networks. These companies include service-oriented firms, such as retail banks, which typically interconnect a few campus locations (including head office, regional sites, and data centers) and many branches, which may number in the hundreds, or even thousands. Other companies that may be affected include manufacturers, which may maintain a number of campus sites, distributed over a metropolitan area network (MAN).

PRICE/PERFORMANCE CHALLENGES
In many campus networks, infrastructures accommodate a mixture of shared media LAN hubs and switches, multi-layer switches, and/or routers — conglomerations that increase complexity and impede scalability. Often the network deployment has been driven by minimizing the price per user, with little consideration for the reliability implications of adding business critical applications, such as telephony. Due to this complexity and the lack of affordable switch and network redundancy, outages are common, resulting in a network that is orders of magnitude less reliable than required for telephony.

While bandwidth is relatively inexpensive in the in-building environment, the unpredictable nature of congestion conditions makes the best-effort handling of packets unacceptable for high-quality telephony. The networking design principles that work very well for TCP/IP data applications are inadequate for IP telephony. Finally, in many enterprises, network management systems are optimized for traditional data traffic rather than for multimedia and reliability-critical operation.

Enterprise users are faced with considerable challenges in managing and scaling these solutions, and introducing consistent quality of service (QoS), security, and policy handling. A key challenge of evolving the campus network infrastructure is that it has to take place in an environment in which the traffic growth is from 30 to 100 percent per year. This growth comes from new applications driven for customer integration (e-commerce, unified customer care), supplier integration (e-business), and e-learning (audio and video streaming and multicast), as well as for productivity and process enhancements (unified messaging, packet telephony, interactive multimedia conferencing).

REQUIREMENTS FOR NEXT-GEN CAMPUS NETWORKS
As they begin deploying new business-critical applications, enterprises will begin placing unprecedented demands on their campus networks. Enterprise requirements placed on the campus networking infrastructure fall into the following categories:

• Scalable performance/bandwidth/ capacity: To accommodate traffic demands both in the LAN and into the WAN.
• Differentiated application networking: To allow networks to meet application and user QoS and security needs without compromising performance.
• Higher network availability: To meet the needs of business for 7x24 operation under normal and congestion conditions.
• Broadband MAN connectivity: To extend campus across the MAN, leveraging the increasing availability of cost-effective fiber-based facilities.
• Simplified network management: Lowering management complexity with resultant reduction in cost of operations.
• Evolution from the installed base toward multi-vendor, standards-based interoperability: To protect past and future investments.
• Lower cost of ownership: Better management of people, bandwidth, and equipment costs.

All of these requirements apply to campus infrastructures of all sizes. That is, they are significant with respect to in-building environments, as well as to WAN/MAN campus infrastructures.

ACHIEVING NEXT-GENERATION CAMPUS INFRASTRUCTURES
The core vision for next-generation campus networks is platform consolidation across in-building and MAN environments, delivering new levels of price/performance and deployment flexibility for telephony, data, and video traffic.
Beyond platform consolidation, key elements of next-generation campus networks include: multigigabit optical networking (meeting traffic growth needs through high-capacity optical networks); policy-enabled networking (meeting application and business QoS and security needs); and system-level reliability (meeting the need for business-critical application networking).

Platform Consolidation
Campus networks today consist of a complex, multi-tiered network. The tiers are as follows:

1. Access or workgroup tier.
2. Campus distribution tier.
3. Campus core tiers.
4. Server aggregation tier.

One alternative to this four-tiered network is a consolidated, high-density, high-capacity, fault-tolerant campus platform. This consolidated platform eliminates the need for a distribution layer between the access and core layers, as well the server aggregation layer between the core layer and server farms. The benefits that result include lower cost, lower latency and higher availability.

The consolidated platform presents a considerably simplified two-tiered campus network topology, consisting of the access/workgroup tier and a core switch tier. (Through integrated server switching, the core switch is connected directly to the server farms.)

Server switching provides three levels of functionality, all geared towards choosing the best available server to handle client requests. The first, and simplest, level of functionality provides balancing and redundancy on a local basis. The second level adds content awareness, allowing, for example, a customer query to be handled differently from a customer order. The third level extends the functions of the previous two levels across geographically dispersed servers and redirects traffic based on server proximity.

The advantages of this new architecture include: growth flexibility and scalability through wirespeed switching; increased network resilience through fully distributed switching, hot-swap modules, and dynamic alternate routing; simplified operations (for example, via single CLI); and reduced cost of sparing.

Enterprises that act on this vision of next-generation campus networks may deploy a common modular system for both workgroup and core tiers, providing a single solution for buildings over an unprecedented size range. Such a platform is configurable as an intelligent workgroup Ethernet switch (at the price of stackables), or as a modular high-capacity campus routing switch.

With this single platform, plug-and-play frame/cell operation will be available. In addition, the customer will have the option of extending campus price/performance across the MAN. Platform consolidation will expand to incorporate switch server functionality, integrated IP telephony call server capabilities, and interworking with wireless LAN systems.

Within in-building environments, the next-generation infrastructure is based on twisted pair to the desktop (complemented by wireless technologies), and single and multimode fiber in the riser, on the backbone links, and to the high-capacity servers. The access tier is based on Layer 2 switches including virtual LAN (VLAN) support with application-aware operation to deliver policy-enabled networking across the network. The core is based on routing switches (including support for common protocols such as IPX) for wirespeed operation in both unicast and multicast modes.

Redundant MultiLink Trunking (MLT) improves Layer 2 bandwidth and resiliency compared to traditional spanning trees Business-critical reliability is provided through switch fault tolerance features and various network-level mechanisms under unified management. Next-generation campus networks provide MAN fiber-based connectivity via ATM, IP on SONET, or directly on fiber running Dense Wavelength Division Multiplexing (DWDM).

Multigigabit Optical Networking
Transmission rates on fiber are doubling every 12 months. Next-generation enterprise infrastructures are taking advantage of these developments to deliver the capacity required by campus in-building and MAN/WAN networking. The next-generation enterprise infrastructures are based on leveraging optical technologies, though the desktop will continue to be based on twisted pair wiring.
In campus/MAN applications, current FDDI and 100Mbit/s campus and MAN links running over dedicated fiber can be upgraded to gigabit Ethernet over distances as high as 50Km. Gigabit Ethernet running MLT is an option available to effectively increase trunk capacity to N Gbit/s (N up to 16). In addition, 10Gbit/s Ethernet will be available, based (for example) on OC192c components from SONET (in much the same way, gigabit Ethernet “borrowed” technology from fiber channel). The next plateau is running Dense Wavelength Division Multiplexing (DWDM) over this fiber. IP on SONET is also an option in extended campus networks.

Policy-Enabled Networking
Providing preferential treatment for certain applications and users is a key emerging requirement, which is provided in next-generation campus infrastructures, through the addition of switch-and network-level QoS and security capabilities. Each switch along the traffic’s path individually ensures that application QoS and security requirements are met, thus enforcing policies defined in the enterprise, and does so without compromising performance.

A structure of network-wide control mechanisms is established to ensure that the “right” applications and end users have access to network resources. This is the role of policy management, an element of Unified Management. Policy management is an implementation of a set of rules or policies, a means of dictating the access to and use of resources on a per user, application, or company basis.

In an ideal world, applications would indicate their QoS requirements. However, most current applications are unable to do so. Therefore, application awareness is built into next-generation campus switches, and into routers and enterprise network switches at the WAN edge level.

New applications may indicate their QoS requirements using standards such as DiffServ (using Type of Service bits in the IP header, IntServ (using RSVP signaling packets), and IEEE802.1p (at Layer 2). Across the network, a broad range of QoS capabilities are provided, including (for example) IEEE802.1p to DiffServ, RSVP to DiffServ, and DiffServ to ATM.

Policy-enabled networking provides an environment in which application performance can be provided while supporting business-driven controls to manage network resources. It ensures that applications such as telephony, e-commerce, SAP, and Web access are given the appropriate treatment. Policy-enabled networking also ensures that the highest availability (even under failure conditions) is provided to business-critical applications; simplifies operations by providing a unified directory environment; and generally lowers the total cost of ownership by making the best use of available bandwidth.

System-Level Reliability
Today’s LAN/router based networks exhibit reliability that is orders of magnitude below the level required to support business-critical data, telephony, and emerging multimedia applications. Network downtime is caused by a wide variety of factors, including loss of power, loss of facilities, network overload, software bugs, and hardware component failures. These problems exist because of deficiencies in the switches and routers, and the lack of management tools to detect pending problems proactively.

Next-generation campus networks encompass a system-level approach at the switch, network, and network management levels, to achieve very high levels of end-to-end availability. At the switch level, power, interface, control, and switching fabric redundancy and hot swappability are already available in many products.

At the network level, resilience is provided at Layer 1 through link redundancy. At Layer 2, mechanisms such as multilink trunking (MLT) and the Virtual Router Redundancy Protocol (VRRP) are used. At Layer 3, resilience is provided through dynamic routing protocols such as OSPF, complemented by Equal Cost MultiPath (ECMP) routing. Next-generation campus networks also leverage application awareness.

Finally, at the network management level, performance and fault management capabilities can significantly enhance network reliability. Remote traffic monitoring are key for effective remote diagnostics.

CONCLUSION
Today’s campus networks lack the requisite reliability and performance to serve as a unified infrastructure for telephony, data, and video. Next-generation campus networks will acquire the necessary attributes, however, through multi-gigabit optical technologies, policy management, and system-level approaches to reliability — all on a single, modular, high-capacity platform for wiring closet and campus backbones.

Tony Rybczynski is director of strategic marketing and technologies for Nortel Networks’ Enterprise Solutions. This business unit offers a full range of enterprise terminal, workgroup, campus, and wide-area unified networks and applications, through direct and indirect channels. For more information, visit the company’s Web site at www.nortelnetworks.com. E-mail questions or comments to the author at tonyryb@nortelnetworks.com.