You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "Arpit Agarwal (JIRA)" <ji...@apache.org> on 2018/10/22 22:52:00 UTC
[jira] [Commented] (HDDS-710) Make Block Commit Sequence (BCS)
opaque to clients
[ https://issues.apache.org/jira/browse/HDDS-710?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16659781#comment-16659781 ]
Arpit Agarwal commented on HDDS-710:
------------------------------------
Another idea proposed by [~anu] was further simplifying the blockIdentifier by completely eliminating the local Block ID.
This gets us significant design simplicity and feels like a very clean approach by making it explicit that blocks cannot be modified, only replaced. It potentially reduces diagnosability. E.g. we cannot easily relate earlier (overwritten) versions of the block.
> Make Block Commit Sequence (BCS) opaque to clients
> --------------------------------------------------
>
> Key: HDDS-710
> URL: https://issues.apache.org/jira/browse/HDDS-710
> Project: Hadoop Distributed Data Store
> Issue Type: Improvement
> Components: Ozone Client
> Reporter: Arpit Agarwal
> Assignee: Arpit Agarwal
> Priority: Major
>
> An immutable block is identified by the following:
> - Container ID
> - Local Block ID
> - BCS (Block Commit Sequence ID)
> All of these values are currently exposed to the client. Instead we can have a composite block ID that hides these details from the client. A first thought is a naive implementation that generates a 192-bit (3x64-bit) block ID.
> Proposed by [~anu] in HDDS-676.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org