[
https://issues.apache.org/jira/browse/TINKERPOP-1852?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16285838#comment-16285838
]
Marc de Lignie commented on TINKERPOP-1852:
-------------------------------------------
One question I have: can the current recipe be updated without providing a
solution for OLAP? I do not feel comfortable in just suggesting that it might
work for SparkGraphComputer without extensive testing that it has any value
over the OLTP case. Also I feel that a proper distributed algo should provide
performance benefits (py parallellization) over an OLTP case for a graph that
still fits in memory. Further, it should be shown that an OLAP algo for
connected components has the same order of magnitude performance as the one in
Apache Spark Graphx.
That being said I would prefer updating the OLTP case and leaving the OLAP case
as work in progress (while removing the old OLAP example).
> Recipe fails for highly meshed connected components
> ---------------------------------------------------
>
> Key: TINKERPOP-1852
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1852
> Project: TinkerPop
> Issue Type: Bug
> Components: documentation
> Affects Versions: 3.3.0, 3.2.6
> Environment: gremlin-console
> Reporter: Marc de Lignie
> Priority: Minor
>
> On the JanusGraph user list the connected-components recipe was shown to fail
> for a completely meshed component with a size of 35. The successive thread
> [https://groups.google.com/forum/#!topic/janusgraph-users/NmwZyag1w2M] shows
> a better solution, both as a mixed groovy-gremlin and a pure gremlin query
> (with the latter having slightly less well scaling behavior because of the
> required path computations).
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
