[
https://issues.apache.org/jira/browse/HBASE-80?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12669246#action_12669246
]
Erik Holstad commented on HBASE-80:
-----------------------------------
Ran tests for the current Implementation of the memory sensitive LruHashMap and
compared them to HashMap and LinkedHashMap:
Adding to LruHashMap
timer 83 for 100000 adds
Size for map 100000
Reading from LRU
timer 133 for 100000 reads
Deleting from LRU
timer 112 for 100000 deletes
Adding to HashMap
timer 39 for 100000 adds
Size for map 100000
Reading from Hash
timer 110 for 100000 reads
Deleting from Hash
timer 79 for 100000 deletes
Adding to LinkedHashMap
timer 55 for 100000 adds
Size for map 100000
Reading from LinkedHash
timer 148 for 100000 reads
Deleting from LinkedHash
timer 102 for 100000 deletes
I think that the number for reads and deletes looks ok, but there seems to be
something weird going on for puts.
Will dig deeper, to find out what.
> [hbase] Add a cache of 'hot' cells
> ----------------------------------
>
> Key: HBASE-80
> URL: https://issues.apache.org/jira/browse/HBASE-80
> Project: Hadoop HBase
> Issue Type: Improvement
> Components: regionserver
> Reporter: stack
> Assignee: Erik Holstad
> Priority: Minor
> Fix For: 0.20.0
>
> Attachments: cache.patch, hbase-80-v1.patch, WR_PE.java
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
