RDStore: In-network resource datastore with distributed processing of resource graph

The delay in comprehending network resources potentially harms precise network management activities, such as root cause analysis and traffic visualization. The delay tends to be remarkable with the growth of the network. This growth basically arises from the number of routers, and the number is easily multiplied by virtualization technology. Hence, the increase in resource data is accelerating, and in future, we will see a fatal limitation of the current management approach, that is, centralized management. To tackle this issue, a decentralized resource management architecture is investigated, and a distributed datastore for in-network resource information sharing, called in-network Resource information DataStore (RDStore), is proposed. In the framework, a simple network modeling based on a resource graph, called the Graph-based network Resource Information (GRI) model, is also proposed for effectively accessing resource data as well as for reducing resource data itself. The demonstration result shows the capability of the proposed accelerated comprehension of network resources, compared with the conventional centralized management approach.