On Mon, Aug 10, 2009 at 3:57 AM, born2fish <tswan...@yahoo.com> wrote:
> > Hi everyone, > > We have a web app that uses embedded solr for better performance. I would advise against it. We use Solr on sites with millions of page views a month on HTTP. With HTTP keep-alives, the overhead of an http request is minimal as compared to the actual search. You have the advantages of replication as well as the option of adding a http cache in front of Solr. Is there a performance problem you're trying to solve by using embedded solr? > Now we are > trying to deploy the app to a clustered environment. My question is: > > 1. Can we configure the embedded solr instances to share the same index on > the network? Yes. It may be very slow though. Best to benchmark it before going to production. Also, you'll need to make sure that only one Solr instance is writing to the index at one time. It is better to have separate indexes. > > 2. If the answer to question 1 is no, can we configure embedded solr > instances to replicate indexes in a master / slave fashion just like normal > web based Solr? > Yes you can use the script based replication. You'd need to expose a way to call commit on your application if you use embedded solr. -- Regards, Shalin Shekhar Mangar.
