[
https://issues.apache.org/jira/browse/AMQ-8344?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jianjiang Mu updated AMQ-8344:
------------------------------
Description:
Hi, I read activemq kahadb soruce code. I find two problems as below:
# Only lfu cache evict policy was provided. If we can provide more cache evict
policy here, such as lru lruk? As we all known, there are a lot of cache ecict
policy.
# Lfu policy provided in source code is not prefect. You can think if the hot
data cached in it first, but after that, hot data was never visited or quered,
they can’t be evicted. Or some not very hot data visited, but because they are
not very hot like hot data, as a result, some of cache was occupied by hot
data. In other world, the cache was not very effective in this situation.
Thanks.
was:
Hi, I read activemq kahadb soruce code. I find two problems as below:
# Only lfu cache evict policy was provided. If we can provide more cache evict
policy here, such as lru lruk? As we all known, there are a lot of cache ecict
policy.
# Lfu policy provided in source code is not prefect. You can think if the hot
data cached in it first, but after that, hot data was never visited or quered,
they can’t be evicted. Or some not very hot data
visited, but because they are not very hot like hot data, as a result, some
of cache was occupied by hot data. In other world, the cache was not very
effective in this situation.
Thanks.
> about kahadb cache evict policy
> -------------------------------
>
> Key: AMQ-8344
> URL: https://issues.apache.org/jira/browse/AMQ-8344
> Project: ActiveMQ
> Issue Type: New Feature
> Components: KahaDB
> Affects Versions: 5.16.3
> Reporter: Jianjiang Mu
> Priority: Minor
> Fix For: 5.16.3
>
>
> Hi, I read activemq kahadb soruce code. I find two problems as below:
> # Only lfu cache evict policy was provided. If we can provide more cache
> evict policy here, such as lru lruk? As we all known, there are a lot of
> cache ecict policy.
> # Lfu policy provided in source code is not prefect. You can think if the
> hot data cached in it first, but after that, hot data was never visited or
> quered, they can’t be evicted. Or some not very hot data visited, but because
> they are not very hot like hot data, as a result, some of cache was occupied
> by hot data. In other world, the cache was not very effective in this
> situation.
> Thanks.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
