In order to increase the efficiency of resource utilization in the network (i.e., traffic engineering), as well as to provide services free of interruptions, the future DOE terabit-capacity network will need to operate across multiple layers. With the significant increase in available bandwidth, the task of providing services resilient to faults will be particularly critical, as even short interruptions in service will result in very large losses of data at 100Gbps or above per-wavelength rates. To improve the speed of protection or restoration and the efficiency of resources allocated to protection, the resiliency mechanisms will have to operate across the layers as well.
We propose to address these challenges by developing a multi-layer and multi-domain modeling and planning framework to support DOE present needs as well as scale to future requirements. This framework will include:
- A Technology-agnostic, Multidomain, and Multilevel Network Model enabling the development of simple yet efficient cross-layer resource allocation algorithms, optimizing path selection along different objectives (minimizing delay, maximizing availability, minimizing blocking probability, reducing CapEx, minimizing energy, etc.). The framework will interface with PerfSONAR for QoS routing.
- A Multi-layer Service-Specific Protection framework, where high-performance/high availability services can independently specify their desired level of protection. The multi-layer protection both increases reliability and improves the resource efficiency.
- A Dynamic network planning tool providing network performance prediction, resiliency/survivability and what-if analyses that informs equipage deployment decision, path computation algorithm design, and new service definitions.
We plan to leverage SDN technologies and OpenFlow in our strategy to address these problems.
This project is funded by the US Department of Energy.