Naming and name resolution in the future internet: Introducing the NovaGenesis approach

Naming and name resolution have fundamental roles in the current and future Internet architectures. In the contemporary Internet, many limitations and problems on this area are longtime well-known, including the limited number of namespaces, the overloaded semantics of IP addresses and domain names, the use of network layer names at higher-levels of the architecture, and the limited name resolution capabilities of DNS. More recently, information-centric networking (ICN) and ID/Locator splitting have emerged as important approaches for rethinking the role of naming on the Internet. In addition, flat name schemes and innovative name resolution service (NRS) are being proposed to improve content distribution efficiency and provenance. In this paper, we are proposing a novel, hierarchical, distributed, name resolution and network caching service to support any combination of flat or hierarchical, natural language or self-verified naming. We provide a scientific contribution to the problem of structuring and resolving names composed of bindings among any kind of names in hierarchical autonomous domains. Our differentiated ICN approach is the core of a future Internet architecture called NovaGenesis. We evaluate and compare the performance of our approach for hierarchical name resolution services, specially in terms of how much servers are used (utilization). Results suggest that our proposal can be adopted as an alternative to the current Internet or other ICN approaches, like multi-level distributed hash tables. We report a reduction on name resolution services utilization at more than one level of authoritative domains' hierarchy. When the probability of a queried domain name is in the cache of a resolver is equal to 0.547 (the typical value today), NovaGenesis root and second level name resolution servers have experimented less than half of the current DNS servers utilization. Finally, we have experimentally evaluated the performance of our proposal in a local domain using a C++ Linux implementation and Docker™ containers, analyzing how the number of name records and hash table nodes impact on name resolution lookup times.