[
https://issues.apache.org/jira/browse/FLINK-1885?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14499830#comment-14499830
]
ASF GitHub Bot commented on FLINK-1885:
---------------------------------------
Github user markus-h commented on the pull request:
https://github.com/apache/flink/pull/598#issuecomment-93987347
There is no specific usecase, but when you try to process big graphs
locally you often run out of memory with delta iterations.
But the reason I needed this change is a different one. I am doing research
on failure recovery methods in graph analysis. Most Pregel like systems just do
a full checkpointing of all vertices. This was way easier to implement with a
bulk iteration than with delta iterations in Flink so I decided to just provide
gelly with this mode.
> Bulk mode for gelly
> -------------------
>
> Key: FLINK-1885
> URL: https://issues.apache.org/jira/browse/FLINK-1885
> Project: Flink
> Issue Type: Improvement
> Components: Gelly
> Reporter: Markus Holzemer
> Assignee: Markus Holzemer
> Priority: Minor
>
> For my research I need to execute graph algorithm with delta iterations and
> with bulk iterations. So I added an execution mode to gelly that executes the
> vertex centric iteration with a bulk iteration.
> Since gelly normally keeps the whole graph in the solution set of the delta
> iteration, it can only handle graphs that fit into the distributed memory of
> ones nodes (correct me if I am wrong). So a bulk mode could also be usefull
> for handling graphs that do not fit in memory.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
