The Evolving Internet - Traffic, Engineering, and Roles

Abstract:     
Entertainment and real-time applications like voice-over-IP, medical telemetry, network gaming and streaming video are quickly becoming prevalent applications over packet-based communication networks. Not only do these applications demand a very diverse set of network performance requirements, networks themselves are also experiencing rapid growth in the number of users and traffic per user. There have traditionally been two canonical approaches to handling such needs: (i) provide substantially increased bandwidth to create sufficient overall capacity in a best-effort network, or (ii) provide class-based differentiated service to meet each application’s performance requirements. The current network neutrality debate reflects the tension between providing substantial additional capacity to meet the most demanding application needs while exploring viable business models to fund this capacity growth.

We examine three dimensions to lend insight to this ongoing debate. First, we discuss traffic growth projections followed by a review of application performance requirements. Next, while assuming a mix of emerging as well as traditional applications, we develop queueing models for an IP backbone to quantify the economies achievable with a differentiated network compared to a best-effort network with enough capacity to achieve the required performance. Finally, we examine a variety of content delivery models to understand the flexibility needed to achieve the objectives of sustainability and customer expectations.

Importantly, our analysis quantifies the amount of over-provisioning required for an IP backbone that provides best-effort service while still meeting the needs of emerging real-time traffic with delay and loss targets. We calculate the Required Extra Capacity (REC) for a best-effort network to meet the same delay and loss performance for premium class traffic provided by a relatively simple, two-class differentiated-service network. Our results demonstrate that the REC increases as the network utilization increases or as the traffic becomes burstier, or the proportion of premium class traffic decreases. With conservative assumptions about the burstiness of the traffic (2-state MMPP parameters), REC approaches 60% even at average link utilizations of 60%, for a relatively small proportion (e.g., 20%) of premium class traffic.

Various business models have evolved to cover the distribution cost of delivering news and entertainment from content provider to consumers. These fall along a wide spectrum based on their differing proportions of distribution costs that are borne by advertising as compared to that borne by consumer subscription revenue. At one extreme, there are some newspapers whose delivery costs (and in fact all their costs) are entirely funded by advertising revenues and at the other extreme, there are media distribution companies (e.g., satellite radio) that are almost completely funded by the consumer. As the Internet plays a bigger role in the distribution of information and entertainment services, it too as a delivery vehicle will require flexibility so that both network operators and content providers alike can design sustainable pricing models that meet the needs of their customer base. Maintaining this flexibility is critical as the Internet’s role in media and entertainment distribution continues to evolve.

Number of Pages in PDF File: 23

Accepted Paper Series