jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Marcel Reutegger <mreut...@adobe.com>
Subject RE: HDFS with Jackrabbit
Date Thu, 30 Aug 2012 10:22:58 GMT
Hi,

> I can see, you have used HBase in Persistence manager implementation. Do
> you
> really think HBase is better solution for hierarchical data design?

well, the internal PersistenceManager interface in Jackrabbit is rather
designed to store content in a single table. this means HBase is a good
fit, but the problem is transactionality. as you can see the sandbox project
depends on a quite old HBase version and AFAIK the most recent version
does not support transactions anymore.

> I am planning to use the HDFS without HBase and map-reduce functionality
> just because of it provide the capability of replication.
> 
> Can you please put some light on prons and cons using HDFS/HBase for JCR?

AFAIK HDFS with Jackrabbit was never used in production and you'd have to
carefully test it. As mentioned earlier, the only sensible usage I see for
HDFS in Jackrabbit is for the DataStore component.

Regards
 Marcel

Mime
View raw message