On Fri, Dec 12, 2014 at 5:31 PM, Shawn Heisey <apa...@elyograg.org> wrote:
> Using a database view that does the JOIN on the server side is pretty > much guaranteed to have far better performance. Database software is > very good at doing joins efficiently when proper DB indexes are > available ... the dataimport handler is not. > fwiw, now we have zipper to merge ordered results from the DB. Comparing to straightforward N+1 it reduces import time from hours to minutes. https://issues.apache.org/jira/browse/SOLR-4799 -- Sincerely yours Mikhail Khludnev Principal Engineer, Grid Dynamics <http://www.griddynamics.com> <mkhlud...@griddynamics.com>
