Gold Supporters


Workshops‎ > ‎

Workshop on Capacity Sharing (CSWS'13)

Internet data traffic continues to grow rapidly. In future it will not be cost-efficient anymore to extend the network resources in the same ratio than the traffic demand. Instead applications and transport layer mechanism need to seek new technologies to efficiently utilize available network resources, in particular resource allocation and flow management.

In wireless networks, the problem is particularly relevant due to the inherently limited resources, which render a simple "throwing bandwidth at the problem" solution impossible. For example, IP traffic from mobile data is expected to double each year in the next years. The success of powerful mobile devices like smart phones or tablets to access different kind of Internet service is not only promoting the growth in traffic volume but also changing the traffic pattern.

The Internet has always been an open platform for innovation and thus for new services to evolve. Due to cloud computing in large data centers and the growth of mobile devices, the number of interactive applications with tight requirements on delay is increasing. While the Internet, especially the mobile Internet, was mostly engineered to provide a low loss service, low-latency services are not well supported today.

Virtualizing networks and network functions in data centers and carrier networks enables a more flexible and efficient resource utilization - but creates new challenges for sharing capacity among tenants and users of those networks. For example, scaling up network resources can be triggered by observed congestion but can also affect the performance characteristics of virtual network slices. At the same time, Network Function Virtualization and Software-Defined Networking open new possibilities for implementing capacity sharing.

This change in network operation and the strong growth and continual changes in the traffic characteristics and usage behavior raises questions on how to share limited capacity resources fairly and more efficiently. While efficiency is most important when resources are spare, fairness need to be evaluated based on the different quality requirements of the various Internet services that we have today.

In commercially operated networks, traffic management is often used to react to observed congestion or disproportionate resource usage. The challenge is to enable these different kinds of applications and to traffic management that maintains high degrees of resource utilization, a reasonable distribution of capacity without affecting transport protocol and application performance and impeding innovation and scalability. One of the questions that the workshop could address is whether and how Capacity Sharing can support traffic management evolving in this direction. We want to encourage researchers to consider integrated consideration of the problem space over all layers.