Managed Bandwidth

Wireless users are expected to spend $828 million this year on fixed wireless, nearly triple what they spent on the technology in 1999, according to International Data. Currently, wireless technology can offer T1 speed at less than the cost of T1 lines. That leaves open a world of possibilities such as voice and data traveling over the same access line and access to entertainment on that same circuit.

Today’s wireless Internet-service providers (W-ISPs) are attracting new subscribers with the promise of providing high bandwidth and exciting new multimedia content. But the lack of effective management tools within these new environments can result in erratic network performance, slower-than-promised transfer rates and service degradation.

Applications such as streaming video and voice over IP stretch the limits and tend to rip open the seams of corporate networks. The knee-jerk reaction of throwing bandwidth at the Internet connection does not always work; it’s best to manage the bandwidth you already have.

While managing wireless technology, W-ISPs must meet the demands for higher levels of network predictability and availability, as well as provide quality of service (QoS). In order to provide QoS, they are facing growing challenges such as controlling bandwidth monopolization; minimizing network congestion; accurately predicting capacity needs to support customer growth; and containing operation costs. Bandwidth issues aren’t going to wait until the service provider can make sense out of the QoS onslaught.

Although throwing bandwidth at the Internet connection seems like the best solution, in the long run, it has negative repercussions. People and applications tend to use up the bandwidth that’s available. Traffic expands to fit the pipe, and service providers can’t stay ahead of it forever. File transfers will grab as much bandwidth as they can; when that happens, packets get dropped. As applications increasingly demand high bandwidth and low delay, QoS promises to mitigate bandwidth problems by prioritizing traffic so delay-sensitive data gets through first. QoS implementations focus on increasing application uptime by protecting the application from changes in network congestion. Although implementing full-blown, end-to-end QoS might prove daunting at first, guaranteeing application uptime and performance can result in a huge return on investment.

The two best reasons for bandwidth management are control and profitability. Management implies control; service providers ultimately depend on network performance for QoS and availability. If a service provider is tracking bandwidth, then it probably can detect problems before the customer does.

Managed bandwidth equals profitability. Why should a service provider deploy (and maintain) more than necessary? And, if a customer is consuming a large amount of bandwidth, the service provider should be able to charge for it or differentiate service levels and charge for them.

In order for a service provider to manage bandwidth and make a profit, QoS technology must be implemented. QoS technology, or a broadband-services-management product, gives service providers the capability of looking at the bandwidth use of their customers. They then can limit a customer’s bandwidth consumption; gather the necessary information to charge precisely for bandwidth consumed; and deploy the optimal amount of bandwidth. This all leads to the maximum contribution to the bottom line.

Controlling Bandwidth Monopolization
As more Internet users tune in to bandwidth-hungry applications such as music-oriented Web sites, network managers are seeing bandwidth problems caused by downloading and sharing of oversized files.

A problem also exists in universities where students are overloading network resources with excessive traffic. Faculty and administration suffer from reduced available bandwidth and higher response time, making them unable to do their jobs efficiently.

In addition, new multimedia-rich applications such as distance learning and the Web traffic of threaded news groups are demanding a wider bandwidth.

Traffic flow — both inbound and outbound — throughout the distribution network needs to be assessed regularly. If the demand for bandwidth exceeds the assigned capacity at any point, then the broadband-services-management product responds to the congestion using dynamic rate control and dynamic priorities.

Essentially, a broadband-services-management product does not create bandwidth, but provides efficient use of existing bandwidth.

Dynamic rate controls imply lowering the transfer rate, but only at the points of congestion and only for the affected traffic direction. As soon as the lowered rate has the desired effect such as congestion being relieved, the transfer rate is allowed to return to normal. Because traffic conditions are reassessed at regular intervals, the rapid toggling between normal and congested rates makes full use of available bandwidth.

Rate limits need to be defined for both normal and congested conditions. With rate limits in effect even under normal conditions, subscribers experience consistent service. In systems that allow rates to burst when network traffic is light, user expectations are raised to unrealistic levels, leading to frustration when bandwidth competition increases.

Dynamic priorities are a rotation system that works by enabling the service policy to include minimum and maximum priority settings. This rotation system ensures that no user is completely starved of bandwidth during periods of congestion, unless the policies have been deliberately configured to do so.

Network Congestion
Today, more than ever, business profits rely on network performance. New applications are being added to the network, and the result is a mix of critical and non-critical traffic sharing the same resources without management. Peak data rates and bursts in the network traffic are hard to predict; therefore, there’s a need to guarantee performance of mission-critical applications. E-business, enterprise resource planning and real-time applications require performance guarantees. When such applications are added to the network, the need for QoS increases exponentially. The greater the traffic mix, the greater the need to control resources.

A broadband-services-management product maximizes the efficiency of broadband Internet connections by responding dynamically to congestion. To prevent aggressive users from causing prolonged periods of congestion, the product should provide an additional control option such as rate ramps. A rate ramp, triggered by sustained high usage, progressively lowers the transfer rate until usage drops to an acceptable level or until a specified final rate is reached.

An effective broadband-services-management product provides ISPs with real-time and historical insight into service performance, enabling them to predict network capacity; monitoring, reporting and alerting for better service management; an extensive collection of statistics to support capacity analysis and planning; and threshold notification of changes in network performance.

In addition, the product should be able to create unlimited service tiers and management. This enables the ISP to provide an unlimited range of services at distinct price points; gain incremental revenue without purchasing additional bandwidth; define and mange service tiers across all broadband services; and significantly reduce provisioning time over manual methods.

Containing Operation Costs
An opposing view to bandwidth management is to just throw bandwidth at the Internet connection to control network traffic. This is done by providing enough bandwidth to minimize queuing delays, therefore, ensuring minimal latency. Supporters of this view argue that managed bandwidth simply is not necessary since bandwidth is nearly free or becoming so.

By not managing bandwidth, service providers may lack QoS mechanisms and any performance guarantees. Such a process will not suffice for applications that need bounded-network performance to meet worst-case latency or throughput-service requirements.

The main disadvantage is that each user consumes bandwidth without regard to the impact on other users. The end result is bandwidth exhaustion — everyone loses and experiences poor network performance, exactly the situation that exists today with some Internet services during peak traffic periods. Just as building more freeways or adding more lanes appears to generate more traffic, adding more bandwidth to a network only stimulates new applications and more network traffic.

Likewise, in this scenario, excessive bandwidth users get a “free ride,” subsidized by others. In addition, the trend today is toward multiservice — networks that combine voice and video onto the same data highway. In this environment, traffic diversity would seem to demand service differentiation, especially latency control for real-time voice, therefore, increasing the need for bandwidth management.

A broadband-services-management product enables ISPs to reduce provisioning times and costs, and reduces customer-support calls.

Guaranteeing reliable and consistent service remains a problem for W-ISPs. Outdated solutions of throwing bandwidth at the Internet connection no longer work, and are expensive. The ideal solution for an ISP is to find a broadband-services-management product that adds value and profit, while keeping subscribers happy. New traffic-shaping devices promise to prevent bandwidth hogs from overwhelming the provider. The available products allow the user to simply point and click using a Web-based interface to monitor all network traffic. A range of products is available to help W-ISPs deal with the crunch.

Getz (steve.getz@crosskeys.com) is CrossKeys vice president of broadband access products and Dyband software.