[
https://issues.apache.org/jira/browse/HBASE-20894?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16562077#comment-16562077
]
stack commented on HBASE-20894:
-------------------------------
bq. removing the deserializer map would break compatibility between the old and
new formats, right?
Yes sir. To be sure, you'd want to rewrite the deserializer code kept in each
BucketEntry so, you'd need to throw away old persistences.
In the scheme of things, this format, though sub-optimal -- and I think it is
'wrong' because it does not name the deserializer... It just serializes a
'code' so could be problematic if we want to put in place a new data structure
other than hfileblock -- is fine. I could make a patch that just adds doc on
what is going on here with these codings for deserializer and the
UniqueIndexMap indirection since it so baffling?
> Move BucketCache from java serialization to protobuf
> ----------------------------------------------------
>
> Key: HBASE-20894
> URL: https://issues.apache.org/jira/browse/HBASE-20894
> Project: HBase
> Issue Type: Task
> Components: BucketCache
> Affects Versions: 2.0.0
> Reporter: Mike Drob
> Assignee: Mike Drob
> Priority: Major
> Fix For: 3.0.0
>
> Attachments:
> 0001-Write-the-CacheableDeserializerIdManager-index-into-.patch,
> HBASE-20894.WIP-2.patch, HBASE-20894.WIP.patch, HBASE-20894.master.001.patch,
> HBASE-20894.master.002.patch, HBASE-20894.master.003.patch
>
>
> We should use a better serialization format instead of Java Serialization for
> the BucketCache entry persistence.
> Suggested by Chris McCown, who does not appear to have a JIRA account.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
