Description:
Background
Multi-homing and peering practices as well as the demand on more routing flexibility have been contributing to the increasing complexity at routers. Operators expect more router programmability, which further challenges router architectures by inclusion of software-based designs and virtualization of routing as a service. Because of the growing complexity of routing tasks, there is a need for techniques and architectural approaches that reduce or offload complexities on the routers.
Description
Researchers at the University Of Nevada, Reno have developed a new architectural approach, Cloud-Assisted Routing (CAR) that leverages high computation and memory power of cloud services for easing complex routing functions such as forwarding and flow level policy management. A novel property of CAR is its aim to find a middle-ground between a pure local approach that targets to scale router performance and a completely cloud-based approach for seamless and highly flexible routing services. Our technology will significantly benefit the industry as seen with the advantages described below.
Advantages
- Our approach with partial dependence on the cloud resources will allow routers to quickly retake responsibility of delegated tasks in case of cloud-related connectivity or performance problems.
- Because CAR embraces commoditization of router hardware by mitigating complexity from the hardware to the cloud, it can facilitate a healthy transformation to a software-oriented routing market.
- Our method allows for lower cost commodity cloud services to supplement expensive router hardware, allowing for identifying the best routers to peer with.
- CAR allows an array of research opportunities for improving ISP backbones:
- Resiliency to failures via cloud-based forwarding and reroute schemes
- Efficiency via more centralized cloud-based optimizations of intra-domain traffic engineering
- Economic competitiveness via cloud-based on-demand service provisioning
Related Documents
UNR13-008
