There was some discussion about SolrCloud in the opac channel of FOLIO Slack on October 12, 2016. VZG, the head office of the library network GBV in Germany, runs VZG Discovery Service for its member libraries. This service uses VuFind and Lukida for the frontend and GBV Zentral as backend.
GBV Zentral is based on SolrCloud, its search index is split into 10 shards with 3 replicas each, which are distributed across 5 servers. For node management Solrcloud uses Apache ZooKeeper. Rights management is done in the proxy and is implemented using filters, this allows for flexible views (by library, by collection, all libraries of a town, …). This also supports searching for items than can be requested using ILL from some other library of the GBV network. Once a day the data is updated using MARC21 import. GBV Zentral can be used with any discovery layer that supports Solr as search backend (eg. Primo, Blacklight, VuFind, OCLC TouchPoint).
All institutions use the same sorting and indexing in GBV Zentral. Until now there was no need for individual sorting or indexing in the backend. Customization in the frontend was sufficient.
GBV Zentral contains about 161 million records and handles 5 million searches per day.