[ https://issues.apache.org/jira/browse/TINKERPOP-2973?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Stephen Mallette updated TINKERPOP-2973: ---------------------------------------- Component/s: hadoop Affects Version/s: 3.5.6 Issue Type: Improvement (was: New Feature) > GraphComputer should allow vertex properties filtering > ------------------------------------------------------ > > Key: TINKERPOP-2973 > URL: https://issues.apache.org/jira/browse/TINKERPOP-2973 > Project: TinkerPop > Issue Type: Improvement > Components: hadoop > Affects Versions: 3.5.6 > Reporter: Boxuan Li > Priority: Minor > Attachments: image-2023-07-25-22-48-37-510.png, > image-2023-07-25-22-49-16-418.png > > > GraphComputer has vertexFilter and edgeFilter, but not yet a filter for > vertex properties. Many vertex programs don't need many if not all vertex > properties. PageRank, for example, does not need any vertex properties at > all. Having a vertex property filter can greatly reduce memory/disk usage. > > I implemented a vertex property filter and will submit a PR soon. In my use > case, the RDD size reduces from 2.8GB to 2.1GB by applying a vertex property > filter. That's 25% storage saving. > > Before: > !image-2023-07-25-22-49-16-418.png|width=1145,height=46! > After: > !image-2023-07-25-22-48-37-510.png|width=1366,height=28! -- This message was sent by Atlassian Jira (v8.20.10#820010)