Hi Wojtek,
Is this a backup or is it a middleman? I can't say that I have seen
the middleman approach before, but that doesn't mean it won't work.
Are you actually having an issue with the current setup or just trying
to make sure you don't in the future?
-Grant
On Jul 15, 2009, at 1:39 PM, wojtekpia wrote:
I'm building a high load system that will require several search
slaves (at
least 2, but this may grow to 5-10+ in the near future). I plan to
have a
single indexer that replicates to the search slaves. I want indexing
to be
as fast as possible, so I've considered adding another machine
between my
indexer and the search slaves (slave master) to reduce the amount of
work
the indexer has to do (file IO, network traffic to/from slaves). I
will have
frequent updates to my index (every 15 minutes), but they will be
small, and
the index will be optimized nightly. The size of the optimized index
is
about 5GB. Has anyone considered or implemented a solution like
this? Did
you see significant performance improvement on the indexer by doing
it?
Thanks,
Wojtek
--
View this message in context:
http://www.nabble.com/Dedicated-Slave-Master-tp24502657p24502657.html
Sent from the Solr - User mailing list archive at Nabble.com.
--------------------------
Grant Ingersoll
http://www.lucidimagination.com/
Search the Lucene ecosystem (Lucene/Solr/Nutch/Mahout/Tika/Droids)
using Solr/Lucene:
http://www.lucidimagination.com/search
