A scalable mapping approach for locator-identifier-separation architectures with flat node ID space

Locator-Identifier-Separation (LIS) architectures introduce a novel addressing scheme by decoupling the node identification and locator functionalities into separate address parts in order to tackle mobility and multi-homing issues. As part of LIS concepts, a fast and global scalable mapping system is required to resolve the node ID into the corresponding locations. Current mapping systems are typically only applicable on hierarchical node ID spaces that violate the LIS principle, or they suffer from administrative and latency problems. In this paper, we propose a scalable multi-level mapping system that is applicable for flat node ID spaces. We apply hierarchical distributed hash tables (DHTs) and multiple cache levels to provide global scalability, a high grade of administrative autonomy and it reduces potential latency issues that can be caused by non-hierarchical DHTs. Within the scope of our work, we evaluate several hierarchical DHT technologies that can be used for that purpose and estimate the performance of our mapping approach.