Teradata add-on package:

Teradata In-database Graph Analytics

Vantage Graph Engine is a distributed designed specifically to handle large-scale graph data—data made of nodes and relationships. Vantage Graph Engine combines:

• A flexible graph data model
• Utilize Teradata MPP architecture Together, these components allow fast graph traversal, querying, analytics, and real‑time processing.

Why use Vantage Graph Engine: Graph Engines excel where data is heavily interconnected and must be queried quickly. Examples include:

• Finding shortest paths
• Exploring neighborhood relationships
• Running graph neural network workloads
• Performing multi-hop queries
• Real-time analytics

Prerequisites: • Database: o Teradata database V17.20 or above. o A database [GraphDB] for Stored Procedure Installation. o Database Privileges for installer. 1. GRANT CREATE Procedure on [GraphDB] to [Installer]; 2. GRANT EXECUTE Procedure on [GraphDB] to [GraphDB]; 3. GRANT ALL on [TargetDB] to [GraphDB] with grant option; o Database Privileges for user. 1. GRANT EXECUTE Procedure on [GraphDB] to [User]; • Client: o Python 3.8 or above o The following packages are required: 1. teradataml>=20.0.0.9 2. teradatasql>=20.0.0.40 3. pandas>=2.2.0 4. pathlib>=1.0.1 5. plotly>=6.3.1

Installation: • Download and install TeradataGE python package from PyPI: o Command: pip install teradatage • Install Graph Analysis SPs from python with TeradataML and TeradataGE packages:

Teradata python package

from teradataml import * import getpass from TeradataGE import td_graph_function, configure, install_db_objects

Database locator: where graph functions are being installed

configure.graph_install_location = "GraphLib"

Connect to Teradata database

hostname = "xx.xx.xx.xx" user = input(prompt=f"Username for SPs instalation:") password = getpass.getpass(prompt=f"Database password for {user}:") logmech = "TD2" eng = create_context(host = hostname, username = user, password = password, logmech = logmech) print(eng)

Install stored procedures

install_db_objects.install_graph_functions()

Disconnect from database

remove_context()