I don't think using HDFS or HBase will perform for this kind of thing
at all. If you are that large, you should look into distributing your
index into shards and using Solr's distributed search capabilities.
-Grant
On Sep 24, 2009, at 3:25 AM, 梁景明 wrote:
hi, thanks, and now i can index data from hbase to the solr server
using
nutch core.
but the indexdata will be local storage,that 's what i worry
about,to be too
large in local.
MountableHDFS i never use it ,i am not sure weather solr can write
the index
into HDFS,i doubt it
can work without implements Writable in HDFS.
and i think the point is the reading and writing the indexfile in
HDFS just
like it in local filesystem ,
can u make a new index file format witch can use in the HDFS, if it
can ,i
think that will
be a great helpful to distrabuted index.
if solr built on top of lucene , will it be easy to implement the
HDFS file
format?
2009/9/24 Amit Nithian <anith...@gmail.com>
Would FUSE (http://wiki.apache.org/hadoop/MountableHDFS) be of use?
I wonder if you could take the data from HBase and index it into a
Lucene
index stored on HDFS.
2009/9/23 Noble Paul നോബിള് नोब्ळ्
<noble.p...@corp.aol.com>
can hbase be mounted on the filesystem? Solr can only read data
from a
filesystem
On Thu, Sep 24, 2009 at 7:27 AM, 梁景明 <futur...@gmail.com>
wrote:
hi, i use hbase and solr ,now i have a large data need to
index ,it
means
solr-index will be large,
as the data increases,it will be more larger than now.
so solrconfig.xml 's <dataDir>/solrhome/data</dataDir> ,can i
used it
from
api ,and point to my
distrabuted hbase data storage,
and if the index is too large ,will it be slow?
thanks.
--
-----------------------------------------------------
Noble Paul | Principal Engineer| AOL | http://aol.com
--------------------------
Grant Ingersoll
http://www.lucidimagination.com/
Search the Lucene ecosystem (Lucene/Solr/Nutch/Mahout/Tika/Droids)
using Solr/Lucene:
http://www.lucidimagination.com/search