Quality of Service - Part 1: Clinging on to Net Neutrality
Posted 7 August 2011 | 
| 
0 Comments
In the beginning the Internet was designed to transmit packets on a best-effort basis. While the network makes a sincere attempt to transmit packets, there are no guarantees. The network does not allocate resources in advance to ensure reliable or timely delivery. As a result, packets may be delayed or even dropped. It is left to higher layer protocols (such as TCP) to ensure reliable data transfer.
Memory buffers are used to avoid collisions when two (or more) packets arrive at a communications link at the same time. A packet must wait until all the packets ahead of it in the buffer have been processed. It follows, then, that when traffic volumes are high, packet waiting times are also high. Furthermore, as memory buffers are not infinite in size, during periods of heavy traffic buffers become full. The communication system will resolve buffer content by dropping packets.
Traffic management methods are used to balance the trade-off between QoS and network optimality. Traffic management is not just a technical issue, network resources are limited and, like any limited resource, its allocation is subject to economical, political and ethical considerations.
There has been much discussion on retaining the neutral network [1]. In a neutral network, there is parity between users and services in terms of how the network treats their packets. This does not mean there is necessarily fairness between heavy users and light users. Heavy user get a greater share of the resources compared to light users, however, both sets of users bare the burden of congestion equally.
In contrast to the neutral network is the differentiated network. The differentiated network discriminates between users and services. Users are allocated resources according to the QoS requirements (and ability to pay). The aim is either to guarantee or at least optimise QoS metrics, such as delay, jitter, throughput or loss.
The network provider controls the allocation of resources, electing to give some users a better than best-effort service while others get a less than best-effort, in the words of Jacobson [4], the aim of QoS is to:
- "Give better service to some traffic (at the expense of giving worse service to the rest)"
Best-effort services are often deemed to be the absence of Quality of Service. Consider high-priority users allocated resources according to their QoS needs, and any remaining resources are allocated to lower-priority users. One may argue, however, that if a particular set of users is receiving a preferential service relative to another, then any “best-effort” commitment to the low-priority users is not being met. In actual fact, low-priority users are receiving a less than best-effort (LBE) service.
For users of applications that are tolerant to high loss rates, delay and jitter, network providers may offer low-cost LBE services alongside best-effort and high-priority services [2]. LBE packets are delayed or dropped, when the network is congested, in preference to best-effort or high-priority traffic. LBE traffic makes use of off-peak periods when network resources are underutilised.
Despite the advocates of the neutral network, in reality the Internet has lacked neutrality for some time now. Discriminatory practices on the Internet are a reality. Whether it is filtering of spam e-mail or the Chinese government censoring discussions of Taiwanese independence, net neutrality is no longer the issue, rather its about “who discriminates and for what purpose”[5].
The amount of real-time traffic on the Internet is increasing and the network infrastructure needs to be able to support it. Users have a sporadic need for resources, and peak demand is rarely maintained (though often exceeded). It is important to realise that QoS is not about blindly implementing quality-of-service measures (traffic management) but about capacity planning. Traffic management is merely a way of optimising network resources such that excessive over-provisioning can be avoided.
However, in moving to a differentiated services model, we must not lose sight of the fact that the Internet is a public good. Traffic management should not be used to used to solve congestion problems for a priviliged minority. Everyone, regardless of their ability to pay, should receive a base level of service. And this can only be achieved by ensuring the network has sufficient capacity.
References
[1] V G Cerf. U.S. Senate Committee on Commerce, science, and Transportation hearing on “Network Neutrality”, 9 2006. http://commerce.senate.gov/pdf/cerf-020706.pdf.
[2] T Ferrari, T Chown, N Simar, R Sabatino, S Venaas, and S Leinen. Experiments with less than best effort (LBE) Quality of Service, 08 2002. http://www.dante.net/tf-ngn/D9.9-lbe.pdf.
[3] Alan Holt. Network Performance Analysis: using the J Programming Language. Springer, London, UK, 2007.
[4] Van Jacobson. Differentiated Services for the Internet. In Internet2 Joint Applications/Engineering QoS Workshop, 5 1998.
[5] Christian Sandvig. Network neutrality is the new common carrier. The Journal of Policy, Regulation, and Strategy, 7 2006.
by Alan | 7 August 2011
Disclaimer: The opinions expressed in this blog represent those of the blogger or original article writer (if reproduced or linked to), and not those of IP Performance Ltd or IP Performance Ltd's subsidiaries, vendors, distribution channel or other business partners. Furthermore, all content being the responsibility of the individual blogger or original author, it is not intended to malign any religion, ethnic group, club, organisation, company, or individual. All data and information provided on this site is for informational purposes only. IP Performance Ltd makes no representations as to accuracy, completeness, currentness, suitability, or validity of any information on this site and will not be liable for any errors, omissions, or delays in this information or any losses, injuries, or damages arising from its display or use. All information is provided on an as-is basis.
