The new public utility

Want to build and manage your own IP-based network but afraid of the risks involved? Leasing capacity on a national carrier's infrastructure cuts out a lot of the worry - and cost

The reliance on IP-based data networks continues to increase as more carriers clamor to offer broadband services. Service providers eager to deliver new products and services to this growing marketplace face an important strategic business decision: They can build and manage their own IP networks - and incur the associated significant operating costs, technical obsolescence risks and staggering capital investments. Or they can focus on their primary business and turn to the industry's new group of carriers.

These national service providers - which include BroadWing (formerly IXC Communications), Level 3 Communications, Qwest Communications and Williams Communications - have built networks with greater economies of scale and efficiencies and with higher levels of network integrity than can be achieved by any one customer.

In a way, these new networks are analogous to today's public utilities.

Companies don't build their own power plants to generate electricity, drill their own wells for water or explore, produce and deliver gas reserves. They instead gain access to the pipelines built by others. Likewise, customers wanting the most efficient use of IP networks should just plug in, leaving the network transport to the high-fidelity national carriers.

Applications drive the network

The proliferation of IP-based applications such as e-business is good news for the advancement of the Internet industry, but it also pressures ISPs to deliver high-quality and reliable services with peak performance standards and cost-effective pricing structures. As applications such as e-mail, Web hosting and unified messaging become more critical to businesses, so does managing these technologies. Network management has opened up new avenues of servicing commercial organizations and created new opportunities for data service providers. As a result, service providers and enterprise customers rely on these IP networks to deliver the services that create their value in the marketplace.

The term service provider in this context refers to ISPs, application service providers and Internet-centric data centers. Service providers such as PSINet and AboveNet function as service bureaus that deliver portal technology, e-commerce, co-location, specialized applications and Web hosting technologies to enterprise customers and other carriers. Because service providers rely on IP networks as a basic infrastructure to deliver their value-added services, reliable, high-performance, cost-effective IP packet transport is critical for their businesses.

Service providers bring value to their customers and shareholders through the services or applications they sell. However, many have made the decision to build their own Layer 3 IP transport infrastructure by piecing together various parts - leased fiber, private lines, switches, routers, facilities, people, etc. - from multiple carriers and network component vendors instead of buying IP ports on national carrier-based IP infrastructures.

Enterprise business customers also are trying to get into the IP game. These companies, which include Fortune 1000 companies such as Ford Motor Co., Dell Computer and Cisco Systems, differ from service providers because they are direct users of Internet networking technologies, employing these technologies to manage and operate their businesses.

Internet IP networks are implemented within the enterprise for a variety of functions, which can include connecting decentralized business units, delivering in-house intranet applications and linking enterprises with their vendors, partners and customers via extranets. Many corporations today require non-intelligent private line pipes to connect data centers, business units, manufacturing, sales, corporate and enterprise information databases.

Rather than committing to long-term contractual agreements, more companies are looking into building their own networks. However, constructing a national network with private lines also means purchasing network switches, routers and other ancillary components, such as redundant network routes and battery backup, to ensure high availability and reliability. In addition, they also must implement network management applications to measure and manage long-haul performance and hire, train and maintain skilled employees to operate and run the network.

Building their own interconnection IP networks rather than purchasing access on a national carrier-grade IP network puts many providers of IP data networking transport services in a risky position, from an economic and a technological standpoint.

The emergence of new technologies in the areas of optical bandwidth, switching, routing, network partitioning, virtual private networks (VPNs), customized network management and specialized billing services results in carriers who can deliver increased performance with high reliability and scalability at lower costs.

A recent study by The Yankee Group concluded that "operations account for a higher portion of total cost of ownership than ever before, and routers (i.e. intranets) are primarily responsible. Routers consume an inordinate amount of support staff time and also consume the lion's share of capital equipment costs." In short, outsourcing network transport services is more cost-effective.

Specialization of products and services allows service providers and other organizations to focus on what they do best. National IP data carriers have invested billions of dollars in engineering and constructing advanced communications infrastructures that feature the latest technologies and applications. Service providers and enterprise customers need to focus their resources on the products and services being delivered over these networks - whether it is an e-commerce solution from the service provider or a company's corporate interactive training session - rather than on the actual transport network.

Operate or outsource?

To illustrate the benefits and pitfalls of operating a proprietary network vs. outsourcing, the following business case scenario details the hard cost components of building and operating a national next generation IP network and what is associated with leasing a port on a carrier-class national/international IP data network. Hard cost components of building a network include private line transmission, equipment space, training and capital expenditures. Similarly, soft costs, such as technology risks, partnership risks, time to market and control, also need to be taken into account.

The hypothetical situation involves two contrasting business cases. Company A decides to build its own national IP data network to connect 10 geographically dispersed locations (Figure 1). Company B has opted to lease national IP data connectivity to the same 10 locations. Using the same base assumptions listed in the box on this page, the business case analysis compares the build vs. lease dilemma by highlighting the corresponding economic, operational and technical results.

For the purposes of this comparison, each company is a Fortune 1000 corporation headquartered in Dallas with decentralized operations. Each has high bandwidth communications needs that rely on IP technology to deliver data communications, database access, file transfers, various intranet applications and VPNs. Each also is experimenting with voice over IP, IP conferencing and distance learning.

Company A: Constructing a next generation IP network. Company A's management has signed three-year term contracts for DS-3 private lines, which are paid for by the DS-0 mile on a monthly recurring basis. They must purchase all network equipment and on-site customer premises equipment. The client requires on-demand network status at all times and needs to track performance characteristics for efficient management, operation and enhancements.

The figures shown in Table 1 show the associated costs with Company A's business case scenario. The figures shown in both tables exclude non-recurring circuit fees, co-location fees and local loop fees. Point of presence co-location, often used by carrier customers instead of local loops, typically cost $2000 per month recurring and $10,000 per month non-recurring for two racks per site.

Company B: Leasing network services. Company B has decided to lease port connectivity from a large carrier-grade IP data network. As part of this agreement, the carrier provides all long-haul IP transport and network operations center support. Table 1 also shows the associated costs with Company B's business case scenario. Company B can expect the same services it would receive if it had built out its own transport service but without the required economic and technical investments.

The bottom line

Conservatively speaking, there is at least a 40% increase in what a service provider would pay to construct its own IP network - without accounting for ongoing R&D required in future IP technology. Aside from the obvious cost benefits, outsourcing network services satisfies the demand for technological innovation at less risk than if a company implemented these network projects itself.

A flexible outsourcing agreement enables an organization to focus on its core business expertise and dedicate more attention to its strategic direction. With IP networks as the framework for a new concept in public utilities, organizations can take advantage of the ever-changing developments in technology while meeting the demand of the new e-commerce economy - without the burden of maintaining an IT infrastructure and allocating bandwidth.

Cost estimates based on a solution from Cisco Systems (August 1999 pricing). Each site includes redundant hardware to support mission-critical applications:

1. High bandwidth availability at 0.999%

2. 24-hour operations and network monitoring

3. Building private IP network with gateway(s) to public Internet

- Three-node core network with seven additional sites. These are configured to minimize the number of circuits required and still provide redundancy in the event of a circuit failure. Some failure modes at certain times of the day could result in network congestion.

- 10-node IP data network - Atlanta, Chicago, Dallas, Denver, Los Angeles, New York, San Diego, San Francisco, Seattle and Washington.

- Bandwidth and connectivity requirements: DS-3 connectivity in each city; all cities must be able to communicate to all other cities.

- Traffic loads are assumed to be variable and evenly distributed from site to site throughout the day with some room for growth.

- Network built to handle failure of any one link by routing around the failed link.

- Point-to-point connections do not exist from every site to every other site (thus larger capacity circuits in the middle of network); some sites share common links.

- Includes hardware (but not monthly recurring cost) for four connections to the Internet.

- Applications include high-speed data transport. These are largely not real time, but the companies are beginning to experiment with real-time applications, including voice over IP, IP conferencing and distance learning.