[
https://issues.apache.org/jira/browse/HBASE-12947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14299030#comment-14299030
]
Andrew Purtell commented on HBASE-12947:
----------------------------------------
How would you handle conflicting concurrent schema updates? While HBase schemas
are in many ways simpler than RDBMS schemas, you might find this paper about
how Google's F1 manages distributed schema change interesting:
http://db.disi.unitn.eu/pages/VLDBProgram/pdf/industry/p764-rae.pdf . We'd have
some of the same implementation challenges. Start by reading section 4.2.
> Replicating DDL statements like create from one cluster to another
> -------------------------------------------------------------------
>
> Key: HBASE-12947
> URL: https://issues.apache.org/jira/browse/HBASE-12947
> Project: HBase
> Issue Type: New Feature
> Components: Replication
> Affects Versions: 2.0.0
> Reporter: Prabhu Joseph
> Priority: Critical
> Fix For: 2.0.0
>
>
> Problem:
> When tables are created dynamically in Hbase cluster, the Replication
> feature can't be used as the new table does not exist in peer cluster. To use
> the replication, we need to create same table in peer cluster also.
> Having API to replicate the create table statement at peer cluster will be
> more helpful in such cases.
> Solution:
> create 'table','cf',replication => true , peerFlag => true
> if peerFlag = true, the table with the column family has to be created at
> peer
> cluster.
> Special cases like enabling replication at peer cluster also for cyclic
> replication has to be considered.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
