Transaction support: If an algorithm changes the graph and then fails, you wish to rollback the partial changes instead of keeping them.Ĭoncurrency control: You want many users to access the same graph. Also, loading a graph does not require any parsing or pre-processing. You could combine the networkx and pickle modules to do that instead, but NetworkDisk saves the graph automatically, avoiding version conflict and allowing a DB user to directly access the graph DB. Persistence requirements and graph sharing: The graph must be saved on disk. Lack of resources: The graph is too large to fit in RAM. There are several motivations to use on-disk graphs rather than in-memory graphs. Is only required for advanced usage and performance tuning. No knowledge of databases is required for simple use of this module. Without worrying about database-specific technology or learning a new database language. The target audience for NetworkDisk are users of NetworkX that want to manipulate on-disk graphs Some will be fast enough to be used as-is. Some algorithms are not suited for on-disk graphs and will thus have poor performance.
Disk graph full#
We aim to achieve full retro-compatibility with NetworkX core methods to ensure The goal is to be as compatible as possible with (Di)Graph objects of the NetworkX package,īut lifting memory requirements and providing persistence of the Graph. NetworkDisk provides a way to manipulate graphs on disk. NetworkDisk: On-disk graph manipulation
0 kommentar(er)
