April 1999

Net Value - The New Economics Of Networking

BY TONY RYBCZYNSKI

For most firms, the network is not only the infrastructure on which the bits and bytes of everyday business travel; it is also a source of increased revenues and increased productivity. If your network fails, your employees stop working, and your customers and business partners stop interacting with you. Clearly, a network failure can be disastrous. And yet, as dire as network failures may be, network maintenance is subject to cost constraints. In networking, price - no less than performance - is relevant.

To recognize the tradeoff between the mission-critical benefits of the network and the costs, the concept of Network Value, or Net Value, is useful. Briefly, Net Value can be thought of as a ratio, the benefits of a network divided by the costs of a network. More explicitly, the benefits of the network are the quantifiable business results attributable to the network less the cost of network downtime (that is, the lost business and lost productivity that results when the network is not available). The cost of the network is the total cost of ownership of the network, including equipment, bandwidth, and operations.

The mission of every IT manager is to maximize the Net Value for his enterprise. Further, the IT manager may opt for one or another approach depending on the unique demands of a particular network. A LAN/campus environment, for example, may present constraints different from those presented by a WAN environment.

But, regardless of the type of network, and the purposes to which that network will be dedicated, thinking about Net Value allows us to focus on three key strategies:

1. Maximizing application performance. (This strategy may focus on meeting application bandwidth and Quality of Service (QoS) needs.)
2. Minimizing network downtime. (This strategy may focus on meeting the need for business-critical availability.)
3. Minimizing the lifecycle costs of operation. (This strategy should be pursued across multiple dimensions, including people, bandwidth, and capital.)

MAXIMIZING APPLICATION PERFORMANCE
The performance requirements of traditional applications are relatively well understood: three-second response times for mainframe applications; low latency for high-quality telephony and for high-quality audio and video conferencing; bandwidth for FTP, e-mail, and HTTP apps. In these familiar situations, the network design objective is to meet performance requirements at the lowest cost. Further, pursuing this objective has often resulted in multiple application-specific networks being deployed in enterprises.

With the convergence on IP and LAN infrastructures for a broad range of applications, however, the design objective has totally flipped. Now, the objective is to "maximize the performance of the network for the dollar." New applications, many of which are critical to business success, drive the need for application-optimized networks. These applications include e-commerce and Internet-enabled call centers, distance training, integrated messaging, IP telephony, enterprise resource planning, and multimedia collaboration.

Several of these applications are delay-sensitive (for example, those with a telephony or real-time video component), putting additional requirements on the network. Generally, legacy and new applications need bandwidth; latency bounds under normal operating conditions, as well as congestion and failure conditions; and network availability.

The optimal approach toward bandwidth requirements differs in in-building and wide area networks. In the LAN, over provisioning is often an attractive approach. In the WAN, packet-based integration of all inter-site traffic on ATM, on frame relay, or (ultimately) on IP-based virtual private networks is the approach that provides the best price/performance.

To enable new applications, networks require a range of IP QoS networking capabilities to meet and manage latency and bandwidth requirements. These QoS capabilities allow business-critical applications to receive preferential treatment from the network, while minimizing the complexity of end-to-end user management and security.

A pivotal function that allows management of this environment is policy management, which is an implementation of a set of rules or policies, the idea being to dictate user connectivity and network resource priorities. Policy management includes three fundamental functions: provisioning or configuring of the network switches and routers; enforcement of the provisioned policies; and verification (or auditing) of network operation.

MINIMIZING NETWORK DOWNTIME
Because so many critical applications depend on the network, network downtime can cripple a business. The cost of downtime is the loss of business and loss of productivity that occurs when the network is not available. At one end of the spectrum, the cost of downtime for brokerage operations may be as high as $100,000 per minute.

The cost of downtime also goes beyond the immediate impact of lost revenue and lost productivity. From a customer's or business partner's perspective, not being "open for business" can do incalculable damage to a company's customer satisfaction, reputation, and brand identity. From an internal perspective, employee morale can suffer.

These indirect consequences of network downtime can also have significant future costs associated with them, including loss of marketshare. As networks enable more and more business-critical and customer-centric applications, the cost of downtime will continue to rise.

Network downtime is caused by a wide variety of factors, including loss of power, loss of facilities, network overload, service provider problems, software bugs, or hardware component failures. According to a Merit study of 1,867 end users, the most frequently cited causes of network resource outages are related to network overload (that is, performance overload, peak load, and insufficient bandwidth). The study also suggests that network overload is attributable to insufficient infrastructure, as well as the lack of appropriate enterprise management tools to detect overload conditions proactively, before they become a source of contention within the enterprise.

Techniques employed to ensure network availability differ from one organization's network to the next based on existing network configuration, the mission-critical character of the applications on the network, and the cost of downtime. However, several design techniques are common in highly available networks. These fall into three categories:

1. Switch level redundancy at the line card, switch fabric, common control, and power supply levels.
2. Network level redundancy (for example, link-level redundancy, automatic routing and failure recovery, congestion controls).
3. End-to-end network fault and performance management tools allowing network managers to detect resource problems proactively and to tweak network design for maximum availability.

To eliminate the problems highlighted in the Merit study, the network should have enough capacity to handle critical applications at peak load. In the campus, this is accomplished by using high-performance switching. Across the WAN, where bandwidth is expensive, the solution is often not more bandwidth, but more efficient use of existing bandwidth.

MINIMIZING COSTS OF OPERATIONS
Total cost of ownership (TCO) consists of the cost of equipment, bandwidth, and network operations. In the LAN/campus, bandwidth is inexpensive relative to engineering. Thus, bandwidth is used to eliminate complexity in this environment. Across the WAN, on the other hand, bandwidth is relatively expensive.

A 1998 Yankee Group study of WAN TCO indicates that bandwidth is the biggest WAN cost area, accounting for 44 percent of WAN TCO in 1998 and rising to almost 50 percent in 1999. The same study indicates that the cost of operations accounts for more than 31 pecent of 1998 TCO and is expected to account for 36 percent in 1999. Capital equipment costs, which make up the balance, are the least significant aspect of TCO. However, the choice of equipment does influence how effectively bandwidth is managed. One major requirement of WAN infrastructure is to minimize the bandwidth required across the wide area.

Operations costs include installation, training, troubleshooting, maintenance, and other costs associated with managing users, applications, and equipment on the network. According to a study by Infonetics, network management staff spend approximately 45 percent of their time on LAN management and 31 percent of their time on WAN management. The balance of staff time is spent on programming, systems work, and administrative tasks.

Time spent on LAN management is generally higher due to higher complexity, the daily grind of moves, adds, and changes, and fewer outsourcing options than in the WAN. In the WAN, the Yankee Group indicates that routers take up 59 percent of support staff time. Given the large number of routers used in WANs, this statistic alone is not surprising. However, the Yankee Group did have some troubling findings with respect to router management. In 1999, the percentage of support staff time consumed by routers is expected to increase to 65 percent. The average number of routers a support staff member can manage is declining, while the average cost of a support staff member is increasing.

A reduction in TCO is accomplished by using bandwidth effectively (a function of both facilities and equipment) and managing operations effectively. Bandwidth effectiveness is simply having enough bandwidth to support optimal application performance and doing so at the lowest possible cost. Increased bandwidth effectiveness is achieved through high-performance switching in the LAN/campus and through consolidation of multiple traffic types across the WAN.

Layer 2 workgroup switches and Layer 3 routing switches in the campus backbone, both with application awareness, provide a solution that provides substantially improved price/performance and that can adapt to increasing traffic demands, unpredictable traffic flow, and the priority requirements of mission-critical applications. Enterprise WAN switches consolidate multiple traffic types, including voice, video, IP, and legacy data, for transport over the WAN.

Operational effectiveness involves minimizing the hassle and expense associated with installation, training, troubleshooting, maintenance, and other tasks associated with managing users and applications on the network. In other words, it's about making the network easier to operate. Because it is difficult to manage what you can't see, end-to-end network management tools are the most important factor in achieving operational effectiveness.

An end-to-end capability across the LAN and WAN provides a complete network view and simplifies the network manager's ability to manage mission-critical applications. As networks become unified, the ability to manage telephony (voice) with these same network management tools will increase in importance. The best network management software incorporates Web-based GUIs, distributed data collection tools, configuration management tools, and multi-vendor network management support.

High-performance switching in the campus and the WAN can significantly reduce the high operations costs associated with router management. Switch-level features such as self-testing capability, built-in support for RMON, and 10/100 Mbit/sec auto-sensing ease the management requirements on individual devices. To reduce the headaches of updating routing tables, devices across the network should be self-configurable. To ease the process of managing IP addresses, server solutions exist that enable dynamic IP address management. These solutions utilize the Dynamic Host Configuration Protocol (DHCP). Finally, automatic network failure recovery minimizes recovery time in the event of network failure.

Though not an exhaustive list, the factors highlighted above are important for effectively managing both bandwidth and network operations. Effective bandwidth management lowers the total cost of ownership while ensuring capacity for business-critical applications. Effective operations management lowers the total cost of ownership, increases network uptime, and allows network managers to allocate more time to strategic issues.

FOCUSING ON NET VALUE CAN HELP YOUR TOP AND BOTTOM LINES
Maximizing Net Value through application-optimized performance, business critical availability, and lower cost of operations is the path to improved shareholder and customer value. System-level approaches to leveraging technologies that optimize the LAN/WAN infrastructure for maximum price/performance, that extend security across VPNs and Internet applications for improved business reach, that combine telephony and IP to leverage connectivity and new applications, and that unify network management across multiple domains to simplify operations are key elements of an unifying enterprise networking strategy that maximizes Net Value.

Tony Rybczynski is director of strategic marketing and technologies for Nortel Networks' Enterprise Solutions. This business unit offers a full range of enterprise workgroup, campus, and wide-area unified networks, 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 [email protected].