[
https://issues.apache.org/jira/browse/HAMA-707?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13545654#comment-13545654
]
Edward J. Yoon commented on HAMA-707:
-------------------------------------
{code}
@SuppressWarnings("unchecked")
private Map<V, List<M>> parseMessages(
BSPPeer<Writable, Writable, Writable, Writable, GraphJobMessage> peer)
throws IOException {
GraphJobMessage msg = null;
final Map<V, List<M>> msgMap = new HashMap<V, List<M>>();
{code}
...
It looks like grouping by vertexID is a biggest problem. All the msgMap and
outgoingQueue should be keeped in memory during execute each vertex.compute()
method.
> BSPMessageBundle should be able to encapsulate messages serialized in
> ByteBuffer
> --------------------------------------------------------------------------------
>
> Key: HAMA-707
> URL: https://issues.apache.org/jira/browse/HAMA-707
> Project: Hama
> Issue Type: Improvement
> Reporter: Suraj Menon
> Fix For: 0.6.1
>
>
> BSPMessageBundle reads out all the messages from the queue and creates an in
> memory list before sending it to peer. This does not work when all messages
> could not be kept in memory. In case of spilling queue, where the messages
> are already serialized, avoid complete read out of messages.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
