Re: Modify Standalone solr server to use it application without http request

[Erick Erickson <er...@gmail.com>]

Currently, the page you referenced here:

http://wiki.apache.org/solr/SolrReplication

is the standard way to replicate incremental indexes.
You say your "worried about the extra http". Why?
Do you have any evidence that this would be a problem?
Http isn't inherently inefficient at all, and even if it was,
the slave asking occasionally if the master's index has
changed (say once every 5-10 minutes) isn't a worry.

The actual replication itself isn't costly protocol-wise.

So, just try it. Coming up in Solr 4.0 is "distributed indexing"
which is a whole different approach, but it's not released yet.

Best
Erick



On Mon, Feb 27, 2012 at 2:36 PM, Neel <ne...@aspiresys.com> wrote:
> Hi,
>
> We are already using embedded solr in our application. In production we have
> 3 app servers and each app server has a copy of index of each type. These
> indexes built externally once in a week and replaced.
>
> We now want allow incremental indexing and auto update to other servers
> rather than building indexes externally and replacing.
>
> I see there are few old posts
> {http://wiki.apache.org/solr/SolrCollectionDistributionScripts,
> http://wiki.apache.org/solr/CollectionDistribution}that says about
> distribution using scripts and rsync. For me this solution looks difficult
> follow on.
>
> I see standalone solrs provides replication
> (http://wiki.apache.org/solr/SolrReplication)  using http requests. I like
> this approach, but worried about additional http request from application to
> standalone solr server.
>
> I now want to enhance standalone solr server to use it in directely in my
> application by removing the ui stuff etc and also work replication
> automatically. Can you please provide guidence how this can be done.
>
> Thanks in advance,
> Neel
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/Modify-Standalone-solr-server-to-use-it-application-without-http-request-tp3781826p3781826.html
> Sent from the Solr - User mailing list archive at Nabble.com.

Re: Modify Standalone solr server to use it application without http request

[Neel <ne...@aspiresys.com>]

Hi Erick,

Sorry for confusing you. My concern was consuming Standalone solr server
through my application can lead additional http requests. 

I tried to access standalone server using CommonsHttpSolrServer in app and
found it works same as embedded solr server and no code change needed and
not much difference in performance as well. 

Thanks,
Neel

--
View this message in context: http://lucene.472066.n3.nabble.com/Modify-Standalone-solr-server-to-use-it-in-application-without-http-request-tp3781826p3799917.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: Modify Standalone solr server to use it application without http request

[Erick Erickson <er...@gmail.com>]

I'm really confused here. Your first question seemed to be about http
involved in index replication, which really doesn't seem to be
related to your latest post. Can you start over from the beginning?

Best
Erick

On Thu, Mar 1, 2012 at 9:56 AM, Neel <ne...@aspiresys.com> wrote:
> Hi Erick, Thanks for your post.
>
> We are not directly providing search result from lucene index to user. We
> are processing the lucene search result and adding additional information to
> it by getting from different sources[from other lunce indexes or from
> databases]. So,
> consuming search results from solr standalone server in the form xml/json
> using http;
> process xml/json and convert to entities,
> get addinal information required for user  from different sources and
> construct xml/json for entire data and send for UI layer..
>
> I feel processing XML/JSON  and convert to entities can lead some
> performance degrade in high usage environment.
>
> Thanks,
> Neel
>
>
>
> --
> View this message in context: http://lucene.472066.n3.nabble.com/Modify-Standalone-solr-server-to-use-it-in-application-without-http-request-tp3781826p3790656.html
> Sent from the Solr - User mailing list archive at Nabble.com.

Re: Modify Standalone solr server to use it application without http request

[Neel <ne...@aspiresys.com>]

Hi Erick, Thanks for your post.

We are not directly providing search result from lucene index to user. We
are processing the lucene search result and adding additional information to
it by getting from different sources[from other lunce indexes or from
databases]. So, 
consuming search results from solr standalone server in the form xml/json
using http; 
process xml/json and convert to entities, 
get addinal information required for user  from different sources and 
construct xml/json for entire data and send for UI layer..

I feel processing XML/JSON  and convert to entities can lead some
performance degrade in high usage environment.

Thanks,
Neel



--
View this message in context: http://lucene.472066.n3.nabble.com/Modify-Standalone-solr-server-to-use-it-in-application-without-http-request-tp3781826p3790656.html
Sent from the Solr - User mailing list archive at Nabble.com.