Abstract: A storage manager for offloading graph components to persistent storage for reducing resident memory in a distributed graph processing engine is provided. The storage manager identifies a set of graph components required to execute a graph processing operation on a graph in a graph processing engine of a database system and reserves an amount of memory needed to load the set of graph components into memory. The storage manager loads the set of graph components into memory and initiates execution of the graph processing operation using the set of graph components in memory. The storage manager evicts one or more unused graph components from memory in response to receiving a request to free a requested amount of memory from memory.

Abstract: Techniques to efficiently assign available workers to executing multiple graph queries concurrently on a distributed graph database are disclosed. The techniques comprise a runtime engine assigning multiple workers to executing portions of multiple graph queries, each worker in each assignment asynchronously executing a portion of a graph query within a parallel-while construct that includes return statements at different locations, and the runtime engine reassigning a worker to executing another portion of the same or a different graph query to optimize the overall performance of all workers.

Abstract: Techniques to efficiently assign available workers to executing multiple graph queries concurrently on a distributed graph database are disclosed. The techniques comprise a runtime engine assigning multiple workers to executing portions of multiple graph queries, each worker in each assignment asynchronously executing a portion of a graph query within a parallel-while construct that includes return statements at different locations, and the runtime engine reassigning a worker to executing another portion of the same or a different graph query to optimize the overall performance of all workers.

Abstract: Techniques are described for enabling in-memory execution of any-sized graph data query by utilizing both depth first search (DFS) principles and breadth first search (BFS) principles to control the amount of memory used during query execution. Specifically, threads implementing a graph DBMS switch between a BFS mode of data traversal and a DFS mode of data traversal. For example, when a thread detects that there are less than a configurable threshold number of intermediate results in memory, the thread enters BFS-based traversal techniques to increase the number of intermediate results in memory. When the thread detects that there are at least the configurable threshold number of intermediate results in memory, the thread enters DFS mode to produce final results, which generally works to move the intermediate results that are currently available in memory to final query results, thereby reducing the number of intermediate results in memory.

Abstract: A pattern matching engine interprets a query into a data structure resembling a finite state machine. Vertices in the query pattern are treated as states or stages, while edges connecting them are treated as state transitions or hops. To match the full pattern, the first stage is first matched by applying vertex filters, if any. If the vertex is eligible, its edges that satisfy the edge filters, if any, are followed to move to the neighbors that can potentially produce results, thus progressing to the next stage. This process is repeated; if all stages are matched, then the whole pattern has been matched successfully.

Abstract: Systems and methods for improving evaluation of graph queries through depth first traversals are described herein. In an embodiment, a multi-node system evaluates against graph data a graph query that specifies a particular pattern to match by determining, at a first node of the multi-node system, in a particular instance of evaluating the graph query, that one or more first vertices on the first node match a first portion of the graph query and that a second vertex that is to be evaluated next is stored on a second node separate from the first node. In response to determining that the next vertex to be evaluated is stored on the second node separate from the first node, the first node generates a message to the second node comprising one or more results of the first portion of the graph query based on the one or more first vertices, an identifier of the next vertex, and a current stage of evaluating the graph query.

Abstract: A pattern matching engine interprets a query into a data structure resembling a finite state machine. Vertices in the query pattern are treated as states or stages, while edges connecting them are treated as state transitions or hops. To match the full pattern, the first stage is first matched by applying vertex filters, if any. If the vertex is eligible, its edges that satisfy the edge filters, if any, are followed to move to the neighbors that can potentially produce results, thus progressing to the next stage. This process is repeated; if all stages are matched, then the whole pattern has been matched successfully.

Abstract: Techniques are described for enabling in-memory execution of any-sized graph data query by utilizing both depth first search (DFS) principles and breadth first search (BFS) principles to control the amount of memory used during query execution. Specifically, threads implementing a graph DBMS switch between a BFS mode of data traversal and a DFS mode of data traversal. For example, when a thread detects that there are less than a configurable threshold number of intermediate results in memory, the thread enters BFS-based traversal techniques to increase the number of intermediate results in memory. When the thread detects that there are at least the configurable threshold number of intermediate results in memory, the thread enters DFS mode to produce final results, which generally works to move the intermediate results that are currently available in memory to final query results, thereby reducing the number of intermediate results in memory.

Abstract: Techniques are described herein for space-efficient encoding of label information of property graphs. In an embodiment, an input graph is received. The input graph comprises a plurality of entities and a plurality of label sets. Each entity of said plurality of entities is associated with a label set of the plurality of label sets and each label set of the plurality of label sets comprises zero or more labels of a plurality of labels. A first mapping is generated that maps each label of the plurality of labels to a label code. A second mapping is generated that maps each label integer set of a plurality of label integer sets to a label code. Each label integer set of the plurality of label integer sets corresponds to a label set of the plurality of label sets, wherein each label integer set of the plurality of label integer sets comprises label codes from the first mapping that are mapped to each label included in the corresponding label set.

Abstract: Techniques are described herein for space-efficient encoding of label information of property graphs. In an embodiment, an input graph is received. The input graph comprises a plurality of entities and a plurality of label sets. Each entity of said plurality of entities is associated with a label set of the plurality of label sets and each label set of the plurality of label sets comprises zero or more labels of a plurality of labels. A first mapping is generated that maps each label of the plurality of labels to a label code. A second mapping is generated that maps each label integer set of a plurality of label integer sets to a label code. Each label integer set of the plurality of label integer sets corresponds to a label set of the plurality of label sets, wherein each label integer set of the plurality of label integer sets comprises label codes from the first mapping that are mapped to each label included in the corresponding label set.