Least-cost routing: Trickier than you think

Least-cost routing is the latest trend in network routing. To the carrier's carrier, it is a new way to do business. And when the goal is minimizing cost and maximizing revenue, least-cost routing appears to be the answer - on paper, anyway.

But is it really so easy? This is the question that many new carriers are trying to answer.

Laying out the plans on paper makes it seem relatively simple; implementing it is a whole different issue. There are many considerations that carriers must address - such as how routing rules affect their choice of reseller and what network management's role is - that can make least-cost routing a win or lose prospect.

What is it anyway?

To define least-cost routing, this article will focus on an international gateway network. A carrier referred to as Provider A sets up a backbone network. Provider A establishes agreements with several other carriers or resellers that want to transport international traffic.

As illustrated in Figure 1, these carriers may offer to terminate traffic, originate traffic or both. Therefore, the location of Provider A's switches is immaterial as long as there are resellers off those switches that offer service to the intended destinations at low rates.

The carriers/resellers then send traffic to Provider A, and Provider A routes those calls to a different reseller that is offering the best rate for that destination. Thus, the service is sold at one price and terminated at another price for what should be the maximum profit. The catch is that the reseller's rates are constantly changing, which means the routing must change, too.

These routing/translation changes can make or break the entire concept of least-cost routing, and they make a good network management key to survival. With rates changing daily, it is imperative to closely monitor routing changes to ensure service isn't affected. Depending on the number of destinations supported, there can be hundreds of different translations for routing each call. When rates change, so does everything else. The first question is how to institute these translation changes in a timely manner with minimal impact to network performance.

Manual vs. automated

There are two ways to institute translation changes to routing: manual and automated.

From a resource perspective, a manual process requires more headcount. Therefore, as the network grows, so does this resource requirement. Given that it takes one dedicated person about a day to change 60 destinations across three switches, it is easy to quantify the magnitude of this work over a large network. In a similar sized network with three switches, it would take one person a week to complete routing changes for 300 destinations. Because changes to rates occur constantly - because revenue is the major cornerstone for this concept - so do the translations.

Aside from headcount, the other issue is accuracy. Human error is always a factor when making translation changes, particularly at this high frequency level. When the occasional typo occurs, calls get sent nowhere. In addition, it is possible to overwrite network controls that may need to stay in place. In this type of network, these mistakes can be costly. Not only does it affect service to the customer, but it also costs the company huge amounts of money if calls are routed to the wrong reseller.

For example, if a company has sold calls to Zaire for 25 cents but routed to a reseller charging 50 cents, the company is instantly losing money. That is why good network management, specifically traffic management, is so important in a least-cost routing network.

On the other hand, an automated process can reduce implementation time, thus reducing extra headcount requirements as well as minimizing the risk of typos. However, automation doesn't solve all the potential problems. In many cases, an automated least-cost routing solution involves an Excel-type spreadsheet linked to the switch. Changes made to the spreadsheet are automatically translated into switch language and loaded into the switch.

In theory, this is a good idea, but in practice it can be dangerous for several reasons. First, implementing routing rules into this process can be complex. Although it may be possible to include some of these rules in the software, it is unlikely that the full complexities of any network can be completely accounted for without some human intervention.

Consider this scenario: A routing rule may be in place that stipulates no backhaul, or passing a call back to a switch that has already passed the call through once.

As shown in Figure 2, the cheapest reseller for a destination may not be the appropriate first choice from a network perspective because of the "no-backhaul" routing rule. Based on the way the network is engineered, the cheapest reseller may reside on a switch that has no alternative route in the event of a failure because it cannot backhaul the call to the preceding switch. The first choice would need to be a different reseller - perhaps the next cheapest - because this scheme would provide an alternative route choice. Therefore, engineering rules still need to be applied up-front by the person entering changes into the spreadsheet.

Second, the right personnel must be selected for this task. Depending on who makes the changes, it can cause a lot of headaches for network management personnel. Because of engineering rules that affect the logic of routing, it is never simply a matter of typing in the new information and hitting the enter key. The complexity of routing rules necessitates a highly skilled person with technical knowledge of the network to make those determinations as he or she enters the data.

In addition to having the right personnel, the way a company sets up and runs the software program is equally important. Will you run it in small batches or run all the changes at once? Ideally, small batches seem more logical in case there is an error along the way. There is always the risk that the software might override a temporary change in routing because of a recent network performance issue, although the software should provide safeguards for this.

However, in the event that something does go wrong, it takes less work to "undo" a small batch of about 60 routing changes vs. one with 300. On the other hand, because these smaller batches may have links to other batches, it is possible that a routing change isn't correctly implemented until the next batch is run. That could leave a network open to lost calls during the batch runs. This means that timing is everything, and understanding how these batches work together is essential.

Finally, whether the process is manual or automated, this function should always reside with network management personnel because of its impact on the network. In addition, they need to be instrumental in setting up the routing change process because they know what can go wrong better than anyone else. This leads to an increased role for network management personnel in a least-cost routing network.

The role of network management

The importance of good network management personnel in this kind of network cannot be stressed enough. They are the eyes and ears of the network and will almost always be the first to notice when things go wrong. Although network management has many important aspects, the vital area in a least-cost routing network is traffic management.

Because completing calls equals revenue, this is the place where upper management tends to focus. Traffic management personnel are the first to notice a problem with traffic patterns as they change with the implementation of new routing. It is this familiarity with the traffic flow that can minimize costly mistakes in the network. Therefore, it is important that traffic management personnel be involved with the addition of new resellers, added capacity and most important, all routing changes, including controls applied by network management personnel during a failure.

This position ensures that calls are going to the correct destination as well as being completed. Traffic management staff understand what a good completion percentage is and how the performance varies, not only between carriers, but also between destinations. If calls to London were getting 80% completion on one carrier and after a routing change are now at 50%, they know that something isn't right. It's these subtle differences that need to be acknowledged.

Based on the network management role in this type of network, the necessary headcount must be committed to network operations to ensure that the technical aspects of the business are maximized. But close coordination between those individuals making technical decisions and those making commercial decisions is crucial. Depending on how service providers organize these areas, they can have very different outcomes (Figure 3).

Commercial decisions do not just rest on how carriers make routing changes and at what frequency. The carrier also has difficult decisions to make regarding revenue vs. performance. For example, Reseller B offers 20 cents on calls to Paris and completed 80% of those calls. Then, a few weeks later, Reseller C offers 15 cents on calls to Paris but only ends up completing 60% of the calls.

What does a carrier do? Does a provider want fewer completed calls at a cheaper rate or more completed calls at a slightly higher rate? Of course, it depends on what rate the carrier is selling the service. It also depends on what level of service the provider wants to offer those customers. If they aren't complaining, then the decision may ultimately reside with upper management.

However, a decision such as this also affects how network operations personnel react. A clear policy must be established ahead of time in situations where technical and commercial requirements may be at odds with one another. The network operations center (NOC) is there to make snap decisions about performance. If NOC staff see a significant drop in call completion, they may be inclined to re-route.

However, upper management may not want a re-route, depending on the cost differential between the two carriers terminating calls. Therefore, commercial guidelines should be communicated to NOC personnel early so they have the necessary information available to make decisions based on both technical and commercial aspects of the business. Management will have to intervene sometimes, but that should be the exception, not the rule.

If carriers completely separate the commercial aspects of the business from technical, they may blind network management's view of the real network. They can't expect traffic management staff to flag potential problems if they are not involved in the decision-making process. In addition, a lot of time can be wasted if they are hunting for an error that was in actuality a change made by someone outside the NOC.

As mentioned earlier, functions such as making routing changes should stay within network operations. It's easy to think that sales personnel can institute the changes more quickly than network management because they are the first to know of rate changes. But more than likely, they won't have insight into the network or the daily activities to factor in technical considerations. Even in an automated process, some technical decisions still must be made up-front.

In addition to personnel requirements and policies/processes, network management systems are an essential part of the equation. Network management tools can be a catalyst to great things or a recipe for disaster.

Good network management systems need to be put in place to provide information such as call origination and destination, call volume and call completion. Users must be able to easily customize these systems, such as setting thresholds for flagging network issues.

In addition, the ability to easily access and analyze call detail records can significantly enhance performance management. Visibility into the network makes all the difference in how much feedback these individuals can provide and in their impact on the network's performance.

Once carriers have the right systems and right personnel in place, balancing this process is the key. Least-cost routing is by no means simple to manage or maintain. It takes a great deal of planning and continuous coordination to ensure everything is working. Then, perhaps what has looked good on paper can also look good at the bottom line.