lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gili Nachum <gilinac...@gmail.com>
Subject Re: Too much Lucene code to refactor but I like SolrCloud
Date Wed, 03 Dec 2014 19:08:10 GMT
Hi Bill.

I'm migrating of a Lucene based app to SolrCloud as well. My main
motivation is horizontal scalability.
My backend is compelx, so the migration is not one cut off, but a long
process; Currently I have both Lucene and SolrCloud, indexing to both, and
querying from either of them.

The migration could be easy or complex depending on what features of Lucene
you use:
You'll need to refactor your indexing flow to create SolrDocuments and post
them to SolrCloud over REST instead of using an IndexWriter.
You'll need to refactor your queries flow to create Solr requests instead
of a Lucene Query, post over REST instead of using IndexSearcher, translate
Solr Response instead of TopDocs.
You can dump all of the code that creates/opens the index. SolrCloud does
that for you.


On Wed, Dec 3, 2014 at 3:10 PM, Bill Drake <wjadrake@gmail.com> wrote:

> I have an existing application that includes Lucene code. I want to add
> high availability. From what I have read SolrCloud looks like an effective
> approach. My problem is that there is a lot of Lucene code; out of 100+
> java files in the application more than 20 of them are focused on Lucene
> code.  Refactoring this much code seems very risky.
>
> My thought was to migrate the index from Lucene 35 to Solr/Lucene 4.10 then
> after making sure everything still works I would add in the HA. I looked at
> solrj with the hope that it would look like Lucene but it did not look like
> it would simplify the transition.
>
> So the question is can I leave most of the existing Lucene code and just
> make small changes to get the benefit of HA from SolrCloud. Is there a
> better approach?
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message