Graph traversal operator inside a column store

Patent metadata

Field	Value
Publication number	US-9697254-B2
Application number	US-201514634418-A
Country	US
Kind code	B2
Filing date	Feb 27, 2015
Priority date	Dec 13, 2012
Publication date	Jul 4, 2017
Grant date	Jul 4, 2017

Abstract

A system, computer-implemented method, and a computer-readable storage medium for a data graph traversal are provided. The input parameters for traversing the data graph are received. The data graph having a set of vertices and a set of edges are stored in a column based format in a memory cache of a computer device based on the input parameters is traversed. The traversal generates a set of traversed vertices that are the result of the graph traversal.

First claim

What is claimed is:

1. A computer-implemented method for improved graph traversal in a memory-centric database, comprising:
receiving, by at least one processor, an input parameter, wherein the input parameter includes a set of path step descriptions;
recursively traversing an edge table for a data graph, wherein the edge table is a column oriented table in the memory-centric database, and the traversing identifies a set of traversed vertices and further comprises:
selecting a path step description from the set of path step descriptions;
extracting a path predicate from the selected path step description;
evaluating the extracted path predicate against the edge table, wherein the evaluation generates a set of valid edges, and the set of valid edges reduces a scan range during traversing;
determining a direction for scanning the set of valid edges based on the selected path step description; and
generating the set of traversed vertices by scanning the set of valid edges using the determined direction; and

returning the set of traversed vertices, wherein the set of traversed vertices are a set of vertices visited during the traversing.

2. The method of claim 1 , wherein at least one terminating condition is specified in the selected path step description.

3. The method of claim 1 , wherein at least one terminating condition is met when no vertices were identified during the scanning with a previous path step description.

4. The method of claim 1 , wherein evaluating the extracted path predicate against the edge table further comprises:
generating a set of traversed edges by scanning the set of vertices and the set of valid edges in the edge table;
eliminating the set of traversed edges from the set of the valid edges;
generating a set of visited vertices from the set of the traversed edges; and
merging the set of visited vertices into the set of traversed vertices, wherein the set of traversed vertices includes vertices from multiple path step descriptions.

5. The method of claim 1 , further comprising optimizing the traversing, wherein the optimizing comprises:
identifying edge types in the data graph;
generating a plurality of sub graphs, wherein a sub graph in the plurality of sub graphs is associated with an edge type included in the edge types; and
traversing the plurality of sub graphs in parallel to determine the set of traversed vertices.

6. The method of claim 1 , further comprising optimizing the traversing, wherein the optimizing comprises:
determining a topological ordering for a vertex in a set of vertices in the data graph;
for the vertex, identifying a subset of vertices that includes vertices reachable from the vertex based on the topological ordering; and
rearranging the edges in the edge table based on the subset of vertices, such that the edges associated with the subset of vertices are traversed during the traversing.

7. The method of claim 1 , further comprising optimizing the traversing, wherein the optimizing further comprises:
determining a subset of vertices that includes vertices a predetermined path step away from a vertex; and
storing the determined subset of vertices in a location in memory, such that the subset of vertices and the vertex are retrieved using a request for a single memory block during the traversing.

8. A system, comprising:
a graph traversal operator stored in memory and executing on a processor and causes the processor to:
receive an input parameter, wherein the input parameter includes a set of path step descriptions;
recursively traverse an edge table for a data graph, wherein the edge table is stored in a column oriented table in a memory-centric database, and the traversing identifies a set of traversed vertices and to traverse the edge table the graph traversal operator is further configured to:
selecting a path step description from the set of path step descriptions;
extracting a path predicate from the selected path step description;
evaluate the extracted path predicate against the edge table, wherein the evaluation generates a set of valid edges, and the set of valid edges reduces a scan range during traversing;
determine a direction for scanning the set of valid edges based on the selected path step description; and
generate the set of traversed vertices by scanning the set of valid edges using the determined direction; and

return the set of traversed vertices, wherein the set of traversed vertices are a set of vertices visited during the traversing.

9. The system of claim 8 , wherein at least one terminating condition is specified in the selected path step description.

10. The system of claim 8 , wherein at least one terminating condition is met when no vertices were identified during the scanning with a previous path step description.

11. The system of claim 8 , wherein to evaluate the extracted path predicate against the edge table is further configured to:
generate a set of traversed edges by scanning the set of vertices and the set of valid edges in the edge table;
eliminate the set of traversed edges from the set of the valid edges;
generate a set of visited vertices from the set of the traversed edges; and
merge the set of visited vertices into the set of traversed vertices, wherein the set of traversed vertices includes vertices from multiple path step descriptions.

12. The system of claim 8 , further comprising an optimization module configured to optimize the traversal and configured to:
identify edge types in the data graph;

generate a plurality of sub graphs, wherein a sub graph in the plurality of sub graphs is associated with an edge type included in the edge types; and
traverse the plurality of sub graphs in parallel to determine the set of traversed vertices.

13. The system of claim 8 , further comprising an optimization module configured to optimize the traversal and configured to:
determine a topological ordering for a vertex in a set of vertices in the data graph;
for the vertex, identify a subset of vertices that includes vertices reachable from the vertex based on the topological ordering; and
rearrange the edges in the edge table based on the subset of vertices, such that the edges associated with the subset of vertices are traversed during the traversal.

14. The system of claim 8 , further comprising an optimization module configured to optimize the traversal and configured to:
determine a subset of vertices that includes vertices a predetermined path step away from a vertex; and
store the determined subset of vertices in a location in memory, such that the subset of vertices and the vertex are retrieved using a request for a single memory block during the traversal.

15. A tangible computer-readable device having instructions stored thereon that, when executed by at least one computing device, causes the at least one computing device to perform operations, comprising:
receiving an input parameter, wherein the input parameter includes a set of path step descriptions;
recursively traversing an edge table for a data graph, wherein the edge table is a column oriented table in a memory-centric database, and the traversing identifies a set of traversed vertices and further comprises:
selecting path step description from the set of path step descriptions;
extracting a path predicate from the selected path step description;
evaluating extracted path predicate against the edge table, wherein the evaluation generates a set of valid edges, and the set of valid edges reduces a scan range during traversing;
determining a direction for scanning the set of valid edges based on

Assignees

• Sap Se

Inventors

• Paradies Marcus
• Rudolf Michael
• Bornhoevd Christof
• Lehner Wolfgang

Classifications

• G06F16/24566 Primary

  Recursive queries · CPC title

• G06F16/221 Primary

  Column-oriented storage; Management thereof · CPC title

• G06F16/951

  Indexing; Web crawling techniques · CPC title

• G06F16/9024

  Graphs; Linked lists (G06F16/9027 takes precedence) · CPC title

• G06F17/30958

  Physics · mapped topic