You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by MitchK <mi...@web.de> on 2010/05/16 02:43:11 UTC
Re: Merging Solr Cloud into trunk
Unfortunately I can't try out the code of the newest cloud-code, because I am
only using a local machine.
However, I am really interested in the concepts behind the distributed
search-concept of Solr.
At https://issues.apache.org/jira/browse/SOLR-1873 I can't find much about
that and even the wiki says nothing usable to understand the concept.
Maybe I am explaining in a bad way what I mean:
I am no professional and I never worked on a distributed project, however if
I need to integrate a distributed search-feature in a search-application, I
would choose the following concept:
Lookup-directory for master and slaves that may become a master in future:
A Lucene-Index with all existing terms in the cloud and instead of documents
that are matching those terms I would save the name of the node that
store/index such documents.
Query:
After finding the nodes that got relevant data, Solr should query them.
Features one could integrate here are things like distributed-idf (perhaps
stored and managed by the master).
Response:
During a query, one node should be elected as a master-of-a-query-session
(moaq).
Every node responses its top-score + top-scoring-document's unique-ID and
the number of matches for the current query.
The next step is done on the moag: Ordering the responsed top-scores he is
able to create a range.
He sends this range back to every node and gets those result that are
matching the query and the range.
So the traffic over the network is reduced to a minimum.
Using a moag instead of the master makes scaling easier (I *think* - I got
no experiences on that).
I *think* all this can be done with a new layer on the current Solr-logic -
however it's only a short excerpt of ideas.
But it's not important what I think here or whether this is usable, since I
only want to give an example of what kind of information I am searching for
:). Are there some papers which describe Solr's distributed concept in such
a detail?
Thank you in advance for the information
*and* thank you for improving Solr's quality and sharing it with the
community.
- Mitch
--
View this message in context: http://lucene.472066.n3.nabble.com/Merging-Solr-Cloud-into-trunk-tp706528p820411.html
Sent from the Solr - Dev mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org
Re: Merging Solr Cloud into trunk
Posted by MitchK <mi...@web.de>.
No public ressources, papers or something else???
--
View this message in context: http://lucene.472066.n3.nabble.com/Merging-Solr-Cloud-into-trunk-tp706528p824612.html
Sent from the Solr - Dev mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org