In recent years, telecom operators have been forced to revisit their long-held strategies as lower-margin data traffic has begun overtaking higher-margin voice traffic. VoIP’s fast adoption is further accelerating this trend by, in effect, transforming voice traffic into data traffic.

 

On the consumer front, telecom operators are reacting to these changes by rolling out triple and quadruple play strategies, which consist of selling bundles of voice, data and video—and sometimes mobile communication services.

 

On the enterprise front, carriers are also faced with eroding margins as the cost of bandwidth is declining steadily and their clients migrate toward less costly TCP/IP-based networks such as Multi-Protocol Label Switching virtual private networks.

 

To date, traffic increases have partly offset the price decline, allowing carriers to maintain their overall margins. However, the days of continual bandwidth capacity increases by enterprise customers are nearly over. They are weary of bandwidth upgrades and realize that such moves alone won’t solve all their problems.

 

 

The number one issue for infrastructure managers according to Forrester and IDC surveys is guaranteeing the performance of critical applications running over their wide area networks (WANs). This issue is not going to disappear anytime soon because it is driven by several powerful underlying trends.

 

An increasing percentage of large organizations’ employees are located at branch offices—more than 50 percent today by some estimates—and most organizations cannot afford to have servers at every site. As such, these employees access their company’s critical applications through the WAN. However, these applications weren’t designed to be transported over a WAN, but over a LAN with much lower communication delays and much larger bandwidth.

 

Moreover, application designers are not network specialists and tend to write their software without considering the interplay of different applications competing simultaneously for a shared infrastructure: the WAN.   

 

The renewed trend towards data center consolidation, the convergence of voice, data and video over the same infrastructure, and the constant rise of non-critical TCP/IP traffic are all contributing to exacerbate the problem of application performance over the WAN.

 

 

The carriers’ response to this problem has been so far to offer Classes of Service (CoS). 

CoS addresses the issue of undifferentiated treatment of application flows by assigning different applications to different classes. Flows assigned to a higher class will have a minimum amount of dedicated bandwidth and will spend less time in congested router queues. As a result, the theory goes, higher class traffic will have lower transfer delays and packet losses.

 

While CoS is a step in the right direction to address the application performance issue, it isn’t enough. Indeed, early adopters of CoS soon came to realize that they were still having application performance issues. Why is that?

 

Inherently CoS is a simplistic approach to WAN optimization. It is simplistic because it aggregates into a single class several applications. These applications or multiple sessions of a same application then compete for bandwidth in the same class, leading to congestion and poor application performance.

 

Further, it only tackles the issue of bandwidth allocation. But other issues such as latency, which are not directly bandwidth dependent, can cause application performance problems.

 

And finally, it is simplistic because it is a static approach to optimization. Indeed, the sizes of the classes of service and the mapping of applications in different classes are decided at the onset of the service. However, WANs are living organisms where traffic conditions change quickly with the addition of new users, new applications and new sites. The original CoS configuration hypotheses usually become quickly obsolete; but revisiting CoS parameters is a tedious and risky task. Thus, carriers don’t change the parameters and WANs remain unoptimized.

 

Some of these shortcomings have been partially addressed by multiplying the number of Classes—some carriers offer up to six classes—and creating complex overflow rules among classes. But these solutions tend to further complicate the administration of the service without really addressing the needs of end-users.

 

 

Carriers need a technology that directly and automatically links application performance objectives and WAN behavior. Such a system would adapt in real time to the supply and demand interplay between user activities, application performance objectives and network resources. It would provide thorough information of the WAN’s contribution to application performance and right size bandwidth according to application performance objectives.

 

With such a system in place, carriers could finally provide the holy grail of networking to their clients: an application-aware network that could offer Application Service Level Agreements guaranteeing application performances over the WAN.

 

Such a service would not only respond to the number one customer issue, but would also allow telecom operators to move up the value chain and avoid becoming commodity suppliers of bandwidth.

 

Fortunately, such technology has begun hitting the market. The trend toward managing application performance over the WAN is gathering momentum. According to different estimates and definitions, this is already a $500 million to $1.2 billion market and growing by 25 percent to 30 percent a year.

 

Most of this market is comprised of companies buying WAN optimization/traffic management technologies and deploying them themselves. A recent IDC survey sponsored by Ipanema Technologies showed, however, that 47 percent of network managers would be interested in receiving the benefits of traffic management tools as a service.

 

This is a unique opportunity for telecom service providers, who have a natural legitimacy over WAN issues, to offer a value-added service. If they don’t, IT outsourcers will be happy to oblige.