Long-tail Traffic - Network Performance

Network Performance

In some cases an increase in the Hurst parameter can lead to a reduction in network performance. The extent to which heavy-tailedness degrades network performance is determined by how well congestion control is able to shape source traffic into an on-average constant output stream while conserving information . Congestion control of heavy-tailed traffic is discussed in the following section.

Traffic self-similarity negatively affects primary performance measures such as queue size and packet-loss rate. The queue length distribution of long-tail traffic decays more slowly than with Poisson sources. However, long-range dependence implies nothing about its short-term correlations which affect performance in small buffers . For heavy-tailed traffic, extremely large bursts occur more frequently than with light-tailed traffic . Additionally, aggregating streams of long-tail traffic typically intensifies the self-similarity ("burstiness") rather than smoothing it, compounding the problem .

The graph above right, taken from, presents a queueing performance comparison between traffic streams of varying degrees of self-similarity. Note how the queue size increases with increasing self-similarity of the data, for any given channel utilisation, thus degrading network performance.

In the modern network environment with multimedia and other QoS sensitive traffic streams comprising a growing fraction of network traffic, second order performance measures in the form of “jitter” such as delay variation and packet loss variation are of import to provisioning user specified QoS. Self-similar burstiness is expected to exert a negative influence on second order performance measures .

Packet switching based services, such as the Internet (and other networks that employ IP) are best-effort services, so degraded performance, although undesirable, can be tolerated. However, since the connection is contracted, ATM networks need to keep delays and jitter within negotiated limits .

Self-similar traffic exhibits the persistence of clustering which has a negative impact on network performance.

• With Poisson traffic (found in conventional telephony networks), clustering occurs in the short term but smooths out over the long term.
• With long-tail traffic, the bursty behaviour may itself be bursty, which exacerbates the clustering phenomena, and degrades network performance .

Many aspects of network quality of service depend on coping with traffic peaks that might cause network failures, such as

• Cell/packet loss and queue overflow
• Violation of delay bounds e.g. In video
• Worst cases in statistical multiplexing

Poisson processes are well-behaved because they are stateless, and peak loading is not sustained, so queues do not fill. With long-range order, peaks last longer and have greater impact: the equilibrium shifts for a while .

Due to the increased demands that long-tail traffic places on networks resources, networks need to be carefully provisioned to ensure that quality of service and service level agreements are met. The following subsection deals with the provisioning of standard network resources, and the subsection after that looks at provisioning web servers which carry a significant amount of long-tail traffic.