5.7.6 SQL-Based Property Graph Analytics
In addition to the analytical functions offered by the graph server (PGX), the property graph feature in Oracle Spatial and Graph supports several native, SQL-based property graph analytics.
The benefits of SQL-based analytics are:
-
Easier analysis of larger graphs that do not fit in physical memory
-
Cheaper analysis since no graph data is transferred outside the database
-
Better analysis using the current state of a property graph database
-
Simpler analysis by eliminating the step of synchronizing an in-memory graph with the latest updates from the graph database
However, when a graph (or a subgraph) fits in memory, then running analytics provided by the graph server (PGX) usually provides better performance than using SQL-based analytics.
Because many of the analytics implementation require using intermediate data structures, most SQL- (and PL/SQL-) based analytics APIs have parameters for working tables (wt). A typical flow has the following steps:
-
Prepare the working table or tables.
-
Perform analytics (one or multiple calls).
-
Perform cleanup
The following subtopics provide SQL-based examples of some popular types of property graph analytics.