[jira] [Created] (IGNITE-5093) Better heap usage during exchange on large topologies and cache numbers/partitions.

["Alexei Scherbakov (JIRA)" <ji...@apache.org>]

Alexei Scherbakov created IGNITE-5093:
-----------------------------------------

             Summary: Better heap usage during exchange on large topologies and cache numbers/partitions.
                 Key: IGNITE-5093
                 URL: https://issues.apache.org/jira/browse/IGNITE-5093
             Project: Ignite
          Issue Type: Improvement
          Components: general
    Affects Versions: 1.6
            Reporter: Alexei Scherbakov
            Assignee: Alexei Scherbakov
             Fix For: 2.1


I observed huge heap occupation during exchange on large grid installation including 136 nodes/1k caches.

Example from machine with 64g heap:

{noformat}
 num     #instances         #bytes  class name
----------------------------------------------
   1:       897287977    43069822896  java.util.HashMap$Node
   2:       9273162        14866180592  [Ljava.util.HashMap$Node;
   3:       201282292    4830775008    java.lang.Integer
   4:       6247215         983811096      [Ljava.lang.Object;
   5:       3383402         767741664      [C
   6:       12188              669411952       [B
   7:       9923859         635126976       java.util.HashMap
...
{noformat}

Further investigation had showed the heap is polluted during exchange process, which involves creating many hashmaps occupying large amounts of memory.

Proposal: use other datastructures to help keep heap usage low.





--
This message was sent by Atlassian JIRA
(v6.3.15#6346)