Has the heterogeneity envisioned some decades ago still the same heterogeneity we experience today? the heterogeneity targeted in the beginning focused on co-existence, i.e. the ability to seamlessly connect different network technologies and shield the upper layer protocols and end systems from the details of the underlying technologies and protocols. In contrast, today, we are challenged to make the heterogeneous technology concurrently collaborate. In particular, in the wake of the fixed-mobile convergence, networks suddenly face the challenge to either dynamically choose one of the available technologies or even to concurrently use multiple technologies.
For example, modern cities typically provide multiple wireless access technologies, such as GSM, 3G, WLAN or even WiMAX. All these technologies are concurrently available and modern devices are even equipped with multiple radios to take advantage of the concurrent availability of the heterogeneous technology. The concurrent availability of heterogeneous resources puts forward a set of unprecedented challenges. A first challenge is to decide who controls the resources. Given a modern device with multiple radios that can be used in parallel, some instance has to decide which and how many resources should be used. Should the end system control the resources? Technically, an end-system approach is able to take the entire end-to-end path into account, including the non-wireless access network as well as potentially multiple providers, whereas a network provider only has information about the technology he deployed. Moreover, an end system may also take end system resources into account, such as battery life. Economically, the end user ultimately needs to be informed about the costs of using multiple technologies. On the other hand, to make such decisions, an end system needs information about the network, including the availability of the different technologies as well as the actual resource usage. The second challenge is to maximize and manage the usage of the different resources. The availability of multiple technologies in the new heterogeneity allows us to exploit the features of the different technologies and consider and arrange them according to some specific metrics: short-delay paths may be used for delay-sensitive applications such as VoIP (Voice over IP), high bandwidth paths for bulk traffic. Again, the question of who controls the resources influences the ultimate outcome. Finally, the third challenge is to implement the cooperation. In particular, the end-to- end argument, the layering principle and the economical separation between ISPs and end users prevent an easy information exchange among the different layers that hide the technological diversity as well as among networks and end systems. Thus, from a design perspective, it is far from obvious how such an implementation could be done with the Internet protocol stack. Should layering, cross-layer implementation and intermediate layers be considered harmful, or should they be considered as necessary steps towards efficiency for a future development of the Internet? The challenges raised here are fundamental problems that touch to the very core principles of the Internet design, such as the end-to-end argument. Solutions should therefore be considered only in this entire context. Moreover, it is likely that the road towards solutions will have to consider the tussles raised by the competing and conflicting demands, preferences and needs of the different stakeholders.
For example, modern cities typically provide multiple wireless access technologies, such as GSM, 3G, WLAN or even WiMAX. All these technologies are concurrently available and modern devices are even equipped with multiple radios to take advantage of the concurrent availability of the heterogeneous technology. The concurrent availability of heterogeneous resources puts forward a set of unprecedented challenges. A first challenge is to decide who controls the resources. Given a modern device with multiple radios that can be used in parallel, some instance has to decide which and how many resources should be used. Should the end system control the resources? Technically, an end-system approach is able to take the entire end-to-end path into account, including the non-wireless access network as well as potentially multiple providers, whereas a network provider only has information about the technology he deployed. Moreover, an end system may also take end system resources into account, such as battery life. Economically, the end user ultimately needs to be informed about the costs of using multiple technologies. On the other hand, to make such decisions, an end system needs information about the network, including the availability of the different technologies as well as the actual resource usage. The second challenge is to maximize and manage the usage of the different resources. The availability of multiple technologies in the new heterogeneity allows us to exploit the features of the different technologies and consider and arrange them according to some specific metrics: short-delay paths may be used for delay-sensitive applications such as VoIP (Voice over IP), high bandwidth paths for bulk traffic. Again, the question of who controls the resources influences the ultimate outcome. Finally, the third challenge is to implement the cooperation. In particular, the end-to- end argument, the layering principle and the economical separation between ISPs and end users prevent an easy information exchange among the different layers that hide the technological diversity as well as among networks and end systems. Thus, from a design perspective, it is far from obvious how such an implementation could be done with the Internet protocol stack. Should layering, cross-layer implementation and intermediate layers be considered harmful, or should they be considered as necessary steps towards efficiency for a future development of the Internet? The challenges raised here are fundamental problems that touch to the very core principles of the Internet design, such as the end-to-end argument. Solutions should therefore be considered only in this entire context. Moreover, it is likely that the road towards solutions will have to consider the tussles raised by the competing and conflicting demands, preferences and needs of the different stakeholders.
So could Heterogeneity be playing with Us?
