Solr provides a deployment mechanism called SolrCloud. This is a clustered Solr environment that allows for distributed Solr servers, failover, load balancing, and sharding of index data. SolrCloud is the most complex Solr deployment option because it involes the use of a Zookeeper Quarum for managing cluster state and configuration. For information on SolrCloud beyond what we provide, please see the SolrCloud documentation.
Broadleaf provides support for SolrCloud. In particular, Broadleaf supports core swapping (or collection re-aliasing), collection and alias creation, as well as the normal indexing and searching that is supported in other Solr deployment models). Some additional terminology is important to note as well. With Embedded and Stand-Alone Solr servers, one or more Solr cores are configured. These cores contain a Lucene index and are located on a single machine. With SolrCloud, the concept of a "collection" is introduced. A collection is like a virtual core. Where a core is a physical concept on a single Solr node (or instance), a collection is a virtual concept and can be distributed across multiple Solr nodes. Embedded Solr configuration created 2 Solr cores on the file system, one for searching and one for reindexing. With SolrCloud, we create a primary collection and a reindex collection, and they can be clustered across multiple Solr instances.