Cutting Your Long-Distance Costs by Using Telecom Traffic Studies.

A critical challenge for communications managers and others concerned with the operation of telecommunications networks and the selection of services for small as well as large companies is to know how to choose from more than 200 options in long-distance service available today.

There are the discount switched services provided by a growing number of OCCs (other common carriers) and resellers, in addition to AT&T's Direct Distance Dialing (DDD) and Wide Area Telecommunications Service (WATS). There are also terrestrial and satellite private-line services, which offer network configurations that may include tie lines, foreign exchanges, off-premise extensions and special data-transmission hookups.

How can these services and other factors impacting on network configuration and efficiency be compared so that it becomes evident which one--or which combination--provides the best economy? The cost, availability and performance of these offered services vary widely.

It becomes even harder to reach decisions when the specifics of the user's own requirements are added. These may include the number of locations, transmission distances, network configuration, frequently-called-number patterns, use of owned or leased data networks, degree of tandeming capability and other factors.

The telecom manager needs consider this rang of factors when deciding on which service mix presents the best tradeoffs necessary to meet his or her telecommunications requirements. Of most-immediate concern are issues such as line quality, geographical availability, installation schedules, billing, maintenance and other support functions. And, of course, cost.

To help ensure that a network uses the most-economical mix of services, the telecom manager has available to him an important tool--telecommunications traffic studies. Only by analyzing a company's actual telecommunications usage can integlligent decisions be made about the mix of services on the basis of cost

The prospect for reduced long-distance costs is a compelling incentive for organizations with large traffic volumes and complex networks to conduct traffic studies. But companies with telecom costs as low as $5,000 per month may also uncover significant savings.

Since long-distance traffic usually amounts to 60 to 80 percent of all telecommunications charges, even marginal savings in this area will have substantial impact on the bottom line. The importance of this is further underscored by the fact that savings of as much as 50 percent have been realized by the introduction of the new long-distance carriers.

Re-evaluating Telecom Services

Furthermore, the fact that prices (as well as technical features, avaialability and other factors) are subject to frequent changes is, in and of itself, an important reason for performing a periodic re-evaluation of services.

The objective of traffic studies is to develop a structured data base that will allow communications managers to sort through all the service options for long-distance telephone and data transmissions, compare them, and the come up with the most cost-efficient and practical communications network for the required grade of service.

While equipment selection, for example, can be handled by simple prices/performance analysis, and loop performance can be easily established by engineering, the most cost-efficient mix of long-distance carriers requires more extensive analysis. Figure 1 illustrates the three key steps involved in doing a typical traffic analysis.

Traffic studies are performed for a fee or even free of charge by many of the carriers and some telecommunications consultants, and usually require three to four weeks for completion.

Where many of these studies fall short is in the common practice of reconfigurating the user's network with services offered by the carrier providing the study; and in other studies, by considering only a particular type of service offering, such as switched service, or restricting it to a limited type of facility for ease of comparison. A truly objective study must avoid these pitfalls.

Recognizing Merits of a Hybrid

Recognizing the merits of a hybrid network, consisting of dedicated and metered services via a mix of satellite and terrestrial facilities, certain basic criteria can be used from the start of the study, such as:

* In general, were customer usage over a particular long-haul route has as little as 45 minutes of traffic (equal to 10 or 11 calls) per day, a dedicated circuit is usually most appropriate.

* In order to provide the best grade of service--that is, minimized busy signals--it is desirable to provide for metered service to supplement the dedicated circuits for overflow traffic.

* Network efficiency can be improved by utilizing circuits that carry two-way traffic in high-quality voice or data, interchangeable at no extra cost, at up to 9600 b/s (and even 14,400 b/s).

* Additional savings accrue from using dedicated circuits to extend calls to those areas closer to the remote terminations than to the first point of origin.

The result of the study usually is a formal proposal that details an optimized network configuration customized to best meet all of the user's communications requirements and provide maximum cost efficiency.

Taking Study's First Step

The first step in the study is an analysis of telephone and data transmission during a typical usage period, usually one month. Traffic summaries from key locations within a company's network and to/from some off-net locations are analyzed to establish the major traffic patterns and to see how these results will affect the direction or emphasis of the continued study.

Widely distributed calling patterns over long distances--that is, few calls to or from the same place--suggest that switched or metered services would provide the most cost-effective option. But if a great number of calls are going to and coming from the same locations, particularly between specific sets of cities, private lines should be considered.

This way, the more-detailed study can be focused on a matrix of locations that have sufficient long-distance traffic to warrant consideration of all the different types of services.

This portion of the study--analysis of the flow of traffic--usually is subcontracted to independent firms that specialize in performing computerized processing and sorting operations. The base data they work from can be a typical month's phone bill or traffic summaries--either in the form of magnetic tapes, as provided by AT&T on its DDD and WATS services; or in Station Message Detail Recording (SMDR) format, as generated by certain PBXs; or as printouts from the OCCs.

By analyzing calling and called numbers, number of calls, duration of calls, cost and other factors (such as the number of business hours on the network, call concentration and acceptable blocking probability), the direction and volume of the major traffic patterns can be established. This data, in turn, later becomes the basis for judgments on network configuration and selection of services.

the complexity of the study will depend on the traffic volume and the scope of the network. In the simplest case, a cursory manual sorting of the phone bill may immediately reveal the best options. For larger firms with complex telecommunications networks, a variety of network options may have to be investigated, and many communications suppliers might be utilized to establish the most cost-efficient network.

Developing Total Cost per Leg

The next objective of a study is to develop the total cost, per month and year, for each type of telecommunications service available on the principal legs in the network, as thus far determined.

First, however, the raw data must be adjusted to accurately reflect typical usage. This includes compensation for seasonality of the original data, number of calls blocked because of busy trunk lines (that is, offered versus carried traffic), and such other factors as type of trafic (voice or data), hours of usage per day and relative traffic volume per hour during the business day. All of this will help normalize the requirements to a basic reference level.

In conjunction with these adjustments, the telecom manager must also make some important decisions before a proper study can be made.

Establishing Grade of Service

In the basic design of a network, decisions were made regarding the type of switching and interconnection facilities to be used in order to provide the level of access acceptable to the users. Now, in addition, the grade of service desired must be established, which involved several tradeoffs.

A call attempt that receives a busy-signal response can be dropped, resulting in the need to re-dial at a later time. It also can be reassigned to an alternate facility held for a predetermined period for re-dial prior to being reassigned or dropped, or held in a queue pending an available facility.

Once the telecom manager has decided on one of these options, the percentage of calls to be subjected to the "busy" treatment--that is, blocking probability--and the reference busy hour for such blocking must be selected.

Impacting on Blocking Choice

Impacting on these choices are the network routing patterns, "intelligence" in the switch--that is, ARS (automatic route selection), LCR (lease-cost routing) and others--and alternate routing facilities, including local telephone facilities in the major calling regions. Decisions on future office locations, anticipated growth in network traffic, and specific locations to be designated points of origin for network traffic will further help in arriving at meaningful and long-lasting configuration recommendations.

Following the initial sorting and subsequent adjustments based on the above determinations, a matrix can then be developed showing the two-way traffic measured in hours between each location and all other locations on the company network. Figure 2 on page 42 shows a typical chart of this type.

The heaviest routes can be discerned by analyzing the traffic on each leg and combining the total number of hours in both directions.

Another set of charts can also be developed showing heavy traffic routes from each of the company locations to other cities and regions not on the network. Such locations call for one-way traffic, usually known as foreign exchange (FX) circuits; and where traffic is sufficiently heavy, this can be another important area of savings.

This analysis resutls in a network configuration that in most instances will look somewhat different from the existing network. Figures 3 and 4 show a typical network before and after traffic analysis.

At this point, it may be desirable to summarize the basic considerations that enter into the development of a traffic study:

Summarizing the Considerations

* Sample size: Data for continuous one-month's typical usage provide meaninful criteria on which to base a recommendation. Shorter periods--one week or even one day--can be utilized, but the smaller the period and number of calls, the greater the margin of error and the lower the confidence factor.

* Reference traffic unit: Although the overall period being considered is "typical," the traffic patterns during the business day will vary. For one thing, the number of business hours per day for each company may vary, and in each case, the calling patterns per hour will show fluctuations in intensity.

Conventionally, for traffic theory, reference usage is based on traffic during the busiet hour of the day; usage during that hour can then be defined in terms of Erlangs (hours of use during the busy hour) or in CCS (hundrel call-seconds) of holding time.

Specifying the P-Factor

* Grade of service: Since the traffic intensity will vary over the course of the day, circuit capacity can be designed for anticipated traffic during the less-busy hours, meaning that during the the busier hours some percentage of calls will receive a busy signal.

The maximum number of calls to be blocked during the busy hour can be specified. This is known as the blocking proability, or P. If P equals 0.05, for example, one can expect five calls out of 100 to be blocked.

This P-factor, selected for any given route with its recorded traffic, together with the number of circuits and the chosen traffic-dimensioning formula, will determine the grade of service over that particular route.

* Traffic capacity: The number of circuits required for the stated offered traffic at a given P-factor is prescribed in standard graphic or tabular format depending on the traffic-handling formula selected--that is, Erlang B, lost calls cleared; Poisson formula, lost calls held; or a particular queuing procedure.

Since the usage data is supplied in the form of carried traffic, and the charts give numbers of circuits in terms of offered traffic, the usage data must be supplemented with the number of blocked calls before the charts can be used properly.