[
https://issues.apache.org/jira/browse/HBASE-14496?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vladimir Rodionov updated HBASE-14496:
--------------------------------------
Description:
Delayed compaction feature for ExploringCompactionPolicy allows to specify
maximum compaction delay for newly created store files. Files will be eligible
for compaction only if their age exceeds this delay. This will allow to
preserve new data in a block cache. For most applications, the newer the data
is the more frequently it is accessed. The frequent compactions of a new store
files result in high block cache churn rate and affects read performance and
read latencies badly.
The configuration will be global, per table, per column family.
was:
Delayed compaction feature allows to specify maximum compaction delay for newly
created store files. Files will be eligible for compaction only if their age
exceeds this delay. This will allow to preserve new data in a block cache. For
most applications, the newer the data is the more frequently it is accessed.
The frequent compactions of a new store files result in high block cache churn
rate and affects read performance and read latencies badly.
The configuration will be global, per table, per column family.
> Compaction improvements: Delayed compaction
> -------------------------------------------
>
> Key: HBASE-14496
> URL: https://issues.apache.org/jira/browse/HBASE-14496
> Project: HBase
> Issue Type: New Feature
> Reporter: Vladimir Rodionov
> Assignee: Vladimir Rodionov
> Attachments: HBASE-14496.v1.patch
>
>
> Delayed compaction feature for ExploringCompactionPolicy allows to specify
> maximum compaction delay for newly created store files. Files will be
> eligible for compaction only if their age exceeds this delay. This will allow
> to preserve new data in a block cache. For most applications, the newer the
> data is the more frequently it is accessed. The frequent compactions of a new
> store files result in high block cache churn rate and affects read
> performance and read latencies badly.
> The configuration will be global, per table, per column family.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
