Search Engine on Separate Host
The following illustration shows a multi-host installation where
solr
and the indexer are running in a single,
dedicated search host:
To do this you would need to:
-
Install Tomcat on your search host.
-
Deploy the
indexer
web application supplied with the Content Store on the search host. -
Deploy a Solr instance on the search host. See Install Solr for details.
-
Copy the
solr
configuration files supplied with the Content Store to the search host, making sure to modify the index schema to meet your requirements, as described in Customizing the Index Schema. -
Modify your publication web applications to use the
solr
instance on your search host.
Isolating solr
in this way would ensure that
re-indexing, for example, does not adversely affect response times
on your presentation hosts. However, it would
also make the search host a single point of failure. A more robust
solution would be to have two or more search hosts, and direct
requests to them via a load balancing and/or fail-over service so
that:
-
Requests are evenly distributed between the search hosts
-
If one host fails, requests are re-directed to other hosts
Load balancing/fail-over strategies can be implemented in many different ways using a variety of different standard products and technologies. Exactly how you do this is outside the scope of this manual: the point is that since all the components involved in searching and indexing communicate via standard, stateless HTTP requests, you can do it using standard web management techniques.