Ethernet services

Study Quantifies Ethernet Opex Savings

The results of a study comparing the likely operating costs of providing Ethernet services with those of providing legacy services (private lines, Frame Relay, and ATM) were released today by the Metro Ethernet Forum (MEF).

In a nutshell, the study indicates that carriers stand to make an average opex saving of 23 percent (see MEF Releases Ethernet Opex Study).

If that figure is right, it's very significant because operating expenditures account for a huge proportion of carrier costs. In 2001, U.S. carrier opex totaled $77.1 billion -- 71 percent of revenues, according to Federal Communications Commission (FCC) figures quoted in the MEF study.

But is that 23 percent figure right?

Well, it's an average figure for a kickoff. In real life, the savings might range from a few percent to more than 50 percent, according to Nan Chen, MEF president. Most carriers could expect at least 10 percent savings, he adds.

The 23 percent average comes from a two-part study conducted by Brian Van Steen, principal of PointEast Research LLC, for the MEF.

In the first part of the study, Van Steen worked with 36 service providers (a mix of incumbents, competitive carriers, and wholesale operators -- 14 in North America and 22 in Europe) to collect time and cost information for their Ethernet services and their legacy private line, Frame Relay, and ATM services. This data was for:

  • Initial provisioning of services
  • Service bandwidth upgrades
  • Customer site additions
  • Customer service additions
  • Service monitoring requirements
In the second part of the study, Van Steen developed a model of a typical metro network, based on government statistics, for a medium-sized city with a population of 1 to 2 million and between 50,000 and 80,000 business establishments.

Van Steen made some assumptions about how fast customers would buy services, upgrade bandwidths, add new sites to their network, and order extra services. This resulted in a three-year forecast of service requirements and operational tasks.

By applying the costs he'd got from the 36 service carriers, Van Steen was then able to come up with opex comparisons for point-to-point services (legacy private line and the Ethernet equivalent, called E-Line) and multipoint services (legacy Frame Relay/ATM and the Ethernet equivalent, called E-LAN). These are summarized below:

Table 1:
Year 1 Year 2 Year 3
E-Line $306,304 $1,427,694 $2,688,542
E-LAN $55,865 $431,062 $1,197,521
Total Ethernet $362,169 $1,858,756 $3,886,063
Private Line $373,613 $1,784,658 $3,555,853
Frame Relay/ATM $70,542 $550,378 $1,583,967
Total Legacy $444,155 $2,335,036 $5,139,820
Opex Saving $81,986 $476,280 $1,253,757
Opex Saving - Percent
Source: PointEast Research

Over three years, the cumulative operating costs of providing the Ethernet services comes to $6,106,988, compared to $7,919,011 for legacy services. This equates to a saving of $1,812,023 or 23 percent.

The MEF paper breaks down the overall savings in the following way:

Initial Provisioning: Savings of 13 to 19 percent on E-Line services and 17 to 22 percent on E-LAN services, mainly from reducing the time spent by field technicians in provisioning and testing services. Only "occasional savings" stem from the use of "more automated provisioning systems."

Bandwidth Upgrades: Savings of 66 to 82 percent for E-Line services and 69 to 83 percent for E-LAN services. This is because bandwidth can be adjusted remotely, via a software adjustment at the network operations center (NOC). No hardware changes are necessary, and thus no truck rolls are needed.

Service Additions: Savings similar to those of bandwidth upgrades because truck rolls can be eliminated in most cases.

Site Additions: Savings of "approximatley 30 percent" for the same reasons given for initial provisioning savings.

NOC Functions: Savings of 12 percent from "simplified circuit provisioning and upgrades." The study says there are no savings "as of yet" in operations, administration, maintenance, and provisioning (OAM&P) functions.

In fact, service providers attending Light Reading Live seminars on Ethernet services have suggested the reverse -- that the lack of operations support systems (OSS) in this area can actually cost carriers money. Instances of carriers using labor-intensive methods of keeping records, such as spreadsheets, in the early stages of service rollouts have been cited by delegates (see Carriers Face Ethernet 'Black Hole').

Similarly, service definitions are only now being standardized (see Standardizing Ethernet Services). In the long run, this promises to fuel the automation of service-level agreement monitoring and billing, but in the meantime they're likely to depend on costly, error-prone manual processes. Nan Chen acknowledges this and says it demonstrates the scope for further opex savings in the long run.

It's tough to say whether the MEF study has overlooked issues that might counter-balance predicted opex savings, such as losing revenue from making billing errors, or increasing monitoring and maintenance chores by deploying less-than-carrier-grade equipment. It's also worth bearing in mind the following points when reviewing the results:

First: the results are heavily influenced by assumptions of customer behavior -- that the average enterprise user will upgrade bandwidths and buy additional services on a regular basis. It's too early to say whether customers will actually behave in this way.

Second: there's a question mark over the Frame Relay costs in the study. It compares equivalent bandwidth services, so the costs collected by Van Steen aren't representative of the vast majority of the Frame Relay service market, which is T1/E1 (1.5-/2-Mbit/s) speeds or less.

Third: the study demonstrates that Ethernet cuts opex on a circuit-by-circuit basis, but it doesn't capture the costs of creating complete networks. In reality, users won't move wholesale to Ethernet. They'll migrate step by step, at first shifting big sites onto Ethernet while keeping smaller sites on Frame Relay. Right now, this "service interworking" technology is at an early stage of development, which means many processes are manual.

Fourth: E-Line and E-LAN services can be provided over dedicated or shared bandwidth, the shared option typically involving the provision of services over an Multiprotocol Label Switching (MPLS) backbone. This opens another can of worms on whether the cost of managing large-scale MPLS backbones can be kept under control. As usual, vendors say "no problem," and some carriers think otherwise.

Fifth: carriers typically don't buy equipment on the promise of opex savings, regardless of how attractive it might look on paper.

— Peter Heywood, Founding Editor, Light Reading

Archives of Related Light Reading Webinars:

Sign In