Dealing with the complexities of network traffic management normally involves two main practices: 1) proactive traffic metering and management in the access network, and 2) segregation of premium and best-effort traffic within the aggregation network, now popularly called the "middle mile." If you are an operator planning optimum service and ROI strategies, you'll probably need to consider both.
The aggregation network (middle mile). As a traditional network traffic management practice, segregating aggregation network "middle mile" traffic is well established in almost every regulatory jurisdiction worldwide. In the past, it was accomplished largely with separate infrastructure for business and consumer services. This model may still be viable in some areas of high economic density, but in most markets, operators expect to fully converge their networks to control costs. That raises questions about just how separate middle-mile facilities need to be.
Guidance from regulators here is sparse, but it's clear that mingling service classes could result in appeals for neutral handling of both. Separating the services in VPNs or VLANs and providing handling priority to one or the other is likely to be accepted, though some operators think resource assignments should be static rather than the dynamic form used in most virtual networks. Virtual routing, for example, is increasingly favored for service partitioning.
A more rigorous form of partitioning may also be valuable for ensuring that the quality of service (QoS) requirement for premium middle-mile service transport is actually met. Even prioritizing premium traffic may not fully insulate the network against performance variations caused by sharp changes in Internet traffic levels. And with services like IPTV, some service components in the premium service might be attacked distributed denial of service (DDoS) if they were made addressable on the Internet.
Middle-mile segregation is an example of class-of-service traffic management, where traffic is categorized by service and handled according to the SLA (if there is one) associated with each service by engineering middle-mile facilities to meet the QoS needs. While principles here are evolving (increased interest in virtual router technology is an example of this evolution), the regulatory and business framework for middle-mile traffic segregation is well established.
The access network. Access traffic management is normally aimed more at enforcing customer service agreements on ingress of traffic and at attempting to provide "relative QoS" to more time-sensitive traffic like voice or video if the customer agreements and regulatory policies permit.
Access or per-customer traffic management normally takes one of the following forms:
- Customer usage is tracked (upstream, downstream or both) and measured against a usage cap. Exceeding this cap will result in either an overage payment or suspension/reduction of service until the next service period. The service period is normally a month.
- The customer's traffic is analyzed on a per-application basis to establish which packets are associated with time-sensitive activity, and these are prioritized. This will often require a form of deep packet inspection (DPI) to be effective in actually recognizing the real-time traffic.
- The customer's traffic is analyzed as above, but certain applications (usually P2P are "deprioritized" or even blocked. This form of traffic management is almost certain to be subject to regulatory questions in at least some of the world's markets (in the U.S., it is the subject of the August 2009 FCC/Comcast regulatory appeal).
Access-level network traffic management is normally applied in mobile networks, and it is very likely that some form of access-level traffic management will eventually be applied to broadband wireline services as well. Most operators report today that between 5% and 10% of wireline broadband customers generate about 90% of traffic. These use-intensive customers are particularly likely to congest limited cable upstream links and future wireless broadband cells.
Applications and customers that use excessive bandwidth will reduce performance for users at large, and the alternative of over-provisioning the network to support them isn't economically viable. This form of traffic management is critical to prevent middle-mile congestion where access speeds rise, such as with 4G wireless services and FTTH.
So what's the traffic management and middle-mile solution for network operators?
- First, assume that your network will offer premium services and provide segregated middle-mile facilities to support them.
- Second, assume that you will be metering access traffic in some way, and select a flexible approach that conforms not only to current regulatory requirements but also supports emerging regulatory trends.
- Finally, be wary of any service architectures that mingle Internet traffic with other traffic beyond the customer demarcation.
Traffic management is inevitable in packet networks, and with proper attention to issues and regulations, it can be a major help in sustaining and even growing revenue and margins.
Return to part 1: Network traffic management needs in the access and "middle mile" aggregation network
About the author: Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is the publisher of Netwatcher, a journal addressing advanced telecommunications strategy issues. Check out his SearchTelecom.com networking blog Uncommon Wisdom