---

Acknowledgments

The work was partially supported by the National Science Foundation Grant number CCF 1533795, Division of Computing and Communication Foundations under the XPS program

---

WARNING

THIS PROJECT IS UNDER HEAVY DEVELOPMENT AND ITS NOT PRODUCTION READY.

The Hybrid Graph Datastore/Computational Engine

Features

• Distributed, resilient, and fast dynamic and static graph database.
• Support distributed computation.
• Online queries and traversal.
• Scalable to billions of nodes and edges.
• Easy setup for both cluster and single instance installations.
• User friendly and intuitive interface for graph analysis, fast algorithm processing and visualization.
• Designed for Data Mining and Machine Learning.

Building blocks:

Workflow with TruenoDB:

Architecture:

Components:

1. Gremlin (Apache Tinkerpop) [1]: A graph traversal language for intuitive and easy graph analysis.
2. Web Console: Web Interface for graph processing, analytics, visualization, and database management. Data laboratory that connects directly to the database/processing engine.
3. Trueno Core: Database/Computational Engine Core.
4. Elasticsearch [6]: Elasticsearch is a distributed, RESTful search and analytics engine capable of solving a growing number of use cases. Stores all graph structures.
5. Apache Spark [4]: a fast and general engine for large-scale data processing. Used for Distributed Graph Processing (GraphX [5]).

Storage Schema:

• Each Index is a graph.
• Distributed workload and storage.
• Data replication to increase availability.
• Data sharding support for horizontal scaling.
• Support multiple properties for edges and vertices.
• Analytic results (built-in collection of algorithms) stored in graph structure for easier post-analysis.

Roadmap to alpha version:

Component	Percentage
Core	90%
Web Console	70%
Connectors	40%
Oblivious Computation	70%
Auditable Computation	80%
Graph Compute Engine	70%
Backend Storage / Indexing	90%
Gremlin Tinkerpop Integration	90%
Dynamic Graphs	10%
Tensorflow Integration	05%

---

TruenoDB Interface Features: Trueno relies on a fast graph analytic/visualization UI.

1. Angular Material [8] Based.
2. Pure WebSockets [9], no slow HTTP requests.
3. Sigma.js WebGL [10] rendering for high scalability.
4. Gremlin [11] Language Traversal for graph retrieval.
5. XPath for graph component filtering.
6. Point, drag, click based functionality. No complex coding or preparation.
7. Export, Save, Import graphs.

---

Install

$ npm install -g trueno

Talks

• "Secure Architecture for Graph Database Systems". Invited Talk at JPL/NASA. August 29, 2019.

References:

• [1] https://tinkerpop.apache.org/
• [2] https://www.tensorflow.org/
• [3] https://lucene.apache.org/solr/
• [4] https://spark.apache.org/
• [5] https://spark.apache.org/graphx/
• [6] http://www.scylladb.com/
• [7] https://cassandra.apache.org/
• [8] https://material.angularjs.org/latest/
• [9] https://www.websocket.org/aboutwebsocket.html
• [10] http://sigmajs.org/
• [11] https://github.com/tinkerpop/gremlin/wiki
• [12] https://www.elastic.co/products/elasticsearch
• [13] https://doi.org/10.1007/s11042-018-6940-2

© Victor O. Santos, Servio Palacios, Edgardo Barsallo, Miguel Rivera, Peng Hao, Chih-Hao Fang, Ananth Grama