Something like this could be a question asked by millions of Internet users worldwide, answered by operators of the Internet and other large market players as well as intensely discussed by national telecommunications market regulators and governments with the greatest urgency ('net neutrality').
The cause of the underlying problem and appropriate responses are found worldwide in a de-facto 'industry-standard' business model that has been established over years of development of the public Internet and now culminating in the 'Flat-Rate Dilemma' which brings it to a certain end. For an outsider it might all sound pretty odd , but for the author however it became a challenging reality – the ́Introduction of 'Quality of Service (QoS) in the public Internet.
Due to the current Internet architecture and the way the Internet has been built (approx. 36000 interconnected stand-alone networks identified via their AS') and the underlying business model however, it would not be sufficient to only limit these efforts to the Internet backbone network of a single backbone operator.
This overall idea aims to evaluate a strategy to roll out the QoS feature Europe-wide, across network boundaries of the single operator and consequently to open new opportunities for new products and services developments as well as an infrastructure cost optimisation to the benefit of operators, ISPs, ASPs, Content Providers and finally - the end-user.
This explains why the overall project is not only about the introduction of new product features but also a transformation of an industry, which could be seen in years to come as a ́game changer ́.

Telecommunication companies facing two opposing objectives. From the marketing side, innovative products need to be flexible and quickly, yet safely brought to market. From the production perspective on the other hand, cost reduction is a commercial imperative. A development arising from this conflict is the migration of an existing business model to a new ́venture ́.
According to Xipeng Xiao and Lionel M. Ni, the problem is best described as follows. 'Today’s public Internet only provides Best Effort Service. Traffic is processed as quickly as possible, but there is no guarantee as to timeliness or actual delivery. With the rapid transformation of the Internet into a commercial infrastructure, demands for service quality have rapidly developed'.¹⁾

It is therefore obvious that distinguished service classes are necessary.

• Best Effort Class (price efficient)Best used for “bulk” background applications, i. e.,http, FTP, e-mail, etc.
• Premium Class (loss sensitive) Tailored for business-critical applications that can not tolerate even a small amount of packet loss (e.g., Telnet, ERP).
• Multimedia Class (delay sensitive)Optimized for bursty and real time applications that can cope with a degree of packet loss but are highly delay sensitive, e.g., video conferencing.
• Voice Class/real time (delay and jitter sensitive)Defined for the typical profile of Voice over IP (VoIP) traffic, i.e., relatively regular streams of small data packets. It is most suitable for delivering time and jitter (delay variation) sensitive applications and built using strict priority queuing mechanisms.

What does that translate into and why is the introduction of 'QoS' and its consequences so important? A few facts envisaging the problem are listed below

According to Prof. Dr. Rüdiger Zarnekow of TU-Berlin, the current Internet business model can be described as follows:

• Service quality in the public Internet infrastructure is currently using the so-called “Best Effort” model. In this model, all services provided by the network operator are “as good as possible” (“Best Effort ”) transports.
• Tensile quality commitments for a service can not be given. There is no possibility to differentiate quality requirements or to prioritize individual services. Instead, all services in the network are transported with the same quality.
• Network operators are therefore forced to leave their network capacity deliberately oversized in order to avoid capacity constraints ('Overprovisioning').
  

This is not only from an economic point of view inefficient, but carries the risk that value-added services such as IPTV or high-value cross-enterprise business processes are jeopardised by low-value services such as peer-to-peer file sharing, as both compete for the same transmission capacity and quality differentiation is impossible. For this reason, the Best Effort model is no longer capable, to enable quality-sensitive services required by a large number of users ́²⁾

Coherent to the tremendous traffic growth, prices for the IP-Transit product dropped in such a way that profitability of this business will soon vanish.

International Internet Traffic Growth versus IP Transit Price Erosion, 2007-2010



Source: http://www.telegeography.com/

The mechanism can be seen as a leaky bucket whereas two buckets will be connected in a chain. The size hole in bucket on the WAN edge is designed that there won‘t be more output as can transported on the WAN connection.
On the entrance edge each QoS class has its own bucket, the size of the holes in this buckets is equal or smaller than size of the hole on the WAN edge. There is a control valve on the each of the QoS buckets. The minimum size of this vale is the minimum bandwidth of the QoS class.
When operating normally, each class can burst up to the bandwidth of the WAN-Access (CAR) (Not on the Voice class). As long as there is not more traffic on LAN side as can be transmitted on the WAN site nothing dramatic happened.

However, in congested situations, if there is more traffic coming in on the LAN side as can be transmitted on the LAN side, the control valve will close. In this situation, the traffic will be taken down on each class. Not only on the QoS class witch delivers to much traffic! Traffic will be regulated down to the minimum committed rate of the QoS classes and traffic witch can’t be transmitted will be dropped.

Understanding the technology: The leaky bucket mechanism


Source: Deutsche Telekom, Internal Presentation on QoS

With the introduction of QoS, the established ´best effort´ business model in the wholesale sector will change fundamentally. Furthermore, the ´new´ business model - 'Sending Party Pays (SPP)’ will be introduced gradually – internationally.

QoS business model , user based (P2P) content exchange for OTT (Over The Top) services

Source: Zarneko, ´Business models for broadband QoS implementation - OTT´

Once the technology and consequently the new business model have been introduced, new products and services such as (Cloud computing, SaaS with end-to-end Service Level Agreements and across network boundaries can be developed or significantly improved - a novum!). After the establishment of the new business model, the resource optimisation as an important ´by-product´(to avoid further overprovisioning) of a QoS implementation will be demonstrated.

• http://en.wikipedia.org/wiki/Quality_of_service
• http://www.cisco.com/en/US/docs/internetworking/technology/handbook/QoS.html
• http://www.bricklin.com/qos.htm
• 'Deploying QoS for Cisco IP and Next Generation Networks, The Definitive Guide', by Vinod Joseph, Brett Chapman, Publisher: Morgan Kaufmann, 2009

— Udo Lorenz 2011/04/29 17:19