Built-in Storage Engine

CogDB has a built-in storage engine that natively stores data as triples. Your data is persistent by default, meaning it safely survives process restarts and system crashes.

Data Model: Triples

CogDB stores data as triples. Each triple consists of a subject, a predicate, and an object. This represents a directed edge in the graph.

For example:

alice → follows → bob
bob → status → active

Persistence

When you create a Graph and supply a name, CogDB creates a local directory to manage the data. All inserts are persisted to disk.

from cog.torque import Graph

# Creates or loads the "people" graph from disk
g = Graph("people")

# Each call to put adds a single triple to the graph
g.put("alice", "follows", "bob")

There is no external dependency like Postgres or Neo4j required. Your data lives where your code lives, providing the expressive power of a graph database with the operational simplicity of a local file.

Flush Intervals

For interactive use, CogDB flushes data to disk on every write. For bulk inserts, you can configure the flush_interval or use sync() to maximize write speed.