PGX 20.1.1
Documentation

Data Structures in Distributed Execution Mode

As its shared-memory counterpart, the distributed execution mode offers methods for the users to create, modify and use different data structures, such as sets, sequences or maps. This page documents the supported and unsupported features of such data structures in the distributed execution mode of PGX.

Common Unsupported Features for Distributed Data Structures

All data structures are mutable in the distributed mode (i.e., the distributed mode does not implement the toMutable() and isMutable() methods for data structures).

Sets and Sequences

Both sets and sequences support only vertices or edges as element types. All other types are not supported. This implies that all sets and sequences in the distributed mode must be created from a graph.

Maps

The following types are supported for both key and value type in maps: int, long, float, double, boolean, vertex, edge, local_date, time, timestamp, time_with_timezone and timestamp_with_timezone. Note that string is not supported.

Components Proxies

Components proxies, which are output by partitioning algorithms such as WCC, are fully supported in the distributed execution mode of PGX.

Paths

The distributed mode does not yet support computing a path from a source vertex to a single destination, (i.e., algorithms, such as Dijkstra Algorithms are not supported), but the computation from a single vertex source to all other vertices in the graph is supported (for instance, using the Bellman-Ford Algorithms). Individual paths can then be retrieved after running the chosen algorithm by supplying the destination vertex, as in the shared-memory mode.