Kein Problem :)

Actually you could even use the slave cluster for writes (those changes would 
not get replicated back, though).
The timestamping of HBase makes that all possible and easy.


I am working on Master-Master replication (see HBASE-2195).


-- Lars



________________________________
From: Jens Hartung <[email protected]>
To: [email protected]
Sent: Wednesday, September 7, 2011 1:01 PM
Subject: Re: Question about HBase Cluster Replication

Hallo Lars,

thanks for your response. I had a discussion with my tutor about the 
cluster-replication feature. He thought, this is only for backup 
purposes. And I didn't find the decisive hint, that the slave can be 
used for read-only queries.

Jens

Am 07.09.2011 18:51, schrieb lars hofhansl:
> Hallo Jens,
>
> yes, you can use the slave cluster for read-only queries (but be aware that 
> the replication is asynchronous, which means the slave can be behind).
> Beyond setting up replication there is no other setup needed for this.
>
> We (Salesforce.com) might be adding code to supporting multiple slaves, but 
> it is not clear when we'll get to that, yet.
>
>
> -- Lars
>
>
>
> ________________________________
> From: Jens Hartung<[email protected]>
> To: HBase Mailinglist<[email protected]>
> Sent: Wednesday, September 7, 2011 4:46 AM
> Subject: Question about HBase Cluster Replication
>
> Hello everyone,
>
> I have to evaluate HBase for a university project. Therefore I read the 
> cluster replication document on HBase main site ( 
> http://hbase.apache.org/replication.html ).
>
> According to the missing feature, there can be only one slave cluster 
> currently. Can the slave cluster be used for read only-access (eg. for 
> load-balancing cases)? When yes, how I have to configure the clusters?
>
> Greetings, Jens

Reply via email to