[
https://issues.apache.org/jira/browse/CARBONDATA-484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15770984#comment-15770984 ]
Ben Manes commented on CARBONDATA-484:
--------------------------------------
You might consider using [TinyLFU|
http://highscalability.com/blog/2016/1/25/design-of-a-modern-cache.html] instead of LRU. This can provide significantly higher hit rate by approximating the historic frequencies.
> Implement LRU cache for B-Tree
> -------------------------------
>
> Key: CARBONDATA-484
> URL:
https://issues.apache.org/jira/browse/CARBONDATA-484> Project: CarbonData
> Issue Type: New Feature
> Reporter: Mohammad Shahid Khan
> Assignee: Mohammad Shahid Khan
> Attachments: B-Tree LRU Cache.pdf
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> LRU Cache for B-Tree is proposed to ensure to avoid out memory, when too many number of tables exits and all are not frequently used.
> Problem:
> CarbonData is maintaining two level of B-Tree cache, one at the driver level and another at executor level. Currently CarbonData has the mechanism to invalidate the segments and blocks cache for the invalid table segments, but there is no eviction policy for the unused cached object. So the instance at which complete memory is utilized then the system will not be able to process any new requests.
> Solution:
> In the cache maintained at the driver level and at the executor there must be objects in cache currently not in use. Therefore system should have the mechanism to below mechanism.
> 1. Set the max memory limit till which objects could be hold in the memory.
> 2. When configured memory limit reached then identify the cached objects currently not in use so that the required memory could be freed without impacting the existing process.
> 3. Eviction should be done only till the required memory is not meet.
> For details please refer to attachments.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
Locked
