DROUND: Direct routing on unique network-layer identifier

The naming and routing schemes in traditional TCP/IP were designed in the days when processors were slow and memory was expensive. These schemes are weak in terms of the following aspects: inability to properly support mobility and multi-homing, creating complexity in the network configuration, making network measurement difficult, and so on. To avoid such weaknesses, we propose a new naming and routing scheme called DROUND. DROUND inherits the advantages of the location-identity separation and routing on flat names, but also has several practical advantages of its own. The proposed scheme, which is compatible with TCP/IP, removes the need for rewriting or re-encapsulating the data packet for routing, shortening the time needed for path routing. Although its scaling properties are not ideal, DROUND is feasible for common ISP networks and enterprise networks based on our experiment results. DROUND is valued for consideration to get better network performance.