Article Preview
TopIntroduction
Since the development of early web pages for scientific purposes, the Internet has suffered huge changes mainly due to the improvement of access technologies and the widespread access of users from the home environment. Many services have been designed, developed and deployed, making the web a vast repository of resources for entertainment, research, study, etc. The ideas behind most of these services have been taken from other activities, e.g., e-mail from snail mail, audio conference from telephone services, etc. Audio and video have also taken advantage of this adoption of ideas and, in the 90s, these services were deployed taking concepts from tape and video recorders and TV and radio services. There are two types of audio/video services on the Internet: live-audio/video and audio/video-on-demand. Live audio/video services deliver information just-in-time for the user who is connected. Information could have been previously recorded and stored, or can be captured and broadcast live; in either case, the information can only be seen once. Only one type of interaction is possible, the pause, and its behaviour is similar to that of a TV broadcast. On the other hand, the principles of audio/video-on-demand are totally different. Users request the information at any time and the servers deliver it exclusively. Also, this system allows users to interact with information: pauses, backward and forward jumps are allowed. Its behaviour is similar to that of a videotape or DVD player. Nowadays, most audio/video services are based on streaming technology. This technology permits the distribution of continuous streams of multimedia data which can be reproduced as they are received and without the need of initial storage.
The deployment of live services is very challenging as they usually broadcast live information. Thus, testing is very limited and there is never a second chance to broadcast the contents. This implies that the quality of service must be under control in order to avoid problems in the delivery of contents and achieve maximum user satisfaction. Quality can be guaranteed by reserving network resources, but at the same time increasing the cost: cost and quality need to be balanced (Luna et al., 2003). Thus, aspects such as the selection of the most suitable architecture or the impact of the underlying network technology need to be carefully studied.
In addition, each of the companies involved in the deployment of one of these services may have its own requirements. Thus, the requirements of content providers may differ from those of the users or the network operators. For example, they would like to generate their contents in their own headquarters, or even use different locations for different shows. This situation conditions the architecture of the service and is a challenge for the network infrastructure, as an incorrect configuration may lead to a poor performance and affect the behaviour of other services. Therefore, from a business perspective, it is essential to avoid the risk of losing clients due to poor quality.
To control quality of service, proper engineering techniques need to be used, including the usage of methodologies such as Melendi et al. (2007) or Pañeda et al. (2007). From the detailed analysis of the service to the execution of advanced tests, several steps must be followed to achieve the desired quality of service. In this context, models and workload tools are important elements for service managers. These tools are important to execute tests and draw conclusions about the future behaviour of the service, the qualities which best fit the preferences of the users, the benefits of certain changes in the architecture of the service, the impact of the service on other services running in the same network, the impact of these services on the radio service, the behaviour of the traffic as in Gopal et al. (2006). These conclusions can be used in the coordination between the different companies involved in the business model (Van der Raadt et al., 2005). Furthermore, service models can be used in the development of business models and feasibility analyses for the deployment of new services (Gordijn et al., 2006).