Candidate
Yunling Cui
Title
Gromit - An In-Memory Graph Database
Supervisor
Hiren Patel
Abstract
This work presents the implementation of an in-memory graph database management system called Gromit. This graph database represents large and complex networks using labelled property graphs, and encodes semantic information in property lists of the vertices and edges. Gromit uses a vertex-edge graph model and represent both vertices and edges as entities of the graph. Edges are stored in a doubly linked list manner in main memory. We implement breadth-first traversal and depth-first traversal to retrieve data for queries. This database supports concurrency and implements locking mechanisms for transaction management. We deploy two benchmark suites from social network domain to evaluate our implementation. These are GDBench and LDBC.