TR288 Study of Topology Discovery Service use with Property Graph Databases v1.0.0

This Technical Report documents the working analysis of the use of two popular Property Graph Databases: Neo4j Open Cypher and Apache Tinkerpop/Gremlin, for implementing typical topology based queries. Both analyses are based on a common realistic exemplar model of a set of routers and links in a Paris Data Center. Using the same exemplar was very useful in identifying common features and where there were differences between them.

This study gave the team insights on how to construct the TMF920A Topology Discovery Service Management API Component Suite operations and notifications; so that they can be realized by Property Graph Databases in a simple and direct way, but without the APIs being tightly tied to these Property Graph Database languages. This allows members to use other forms of databases. This turned out to be essential as it turns out that Gremlin is based more on a scripting model, whereas Open Cypher is based on a declarative model, so that simple exposure of each syntax through an API would lead to tight coupling.

The exercise was extremely valuable in understanding how Property Graph Database work for representing topologies and supporting path queries. The team discovered that it was common practice to assign unique ids to both Vertices and Edges and that as a minimum all nodes have a common property – Label/Token, and the ability to add additional properties as required which can differ across each Vertex and Edge.

The Study of the searching mechanism in these Property Graph Databases influenced how the team designed the supporting topology discovery API to meet the requirements in TMF920A.