Measurements from UKWA suggests that sparse faceting on large result sets are markedly slower (x2) than vanilla Solr for some setups. The old option of switching directly to vanilla Solr when the result set exceeds a certain size should be re-introduced until the discrepancy between sparse & vanilla Solr has been eliminated.
Technical note: When the facet.sparse.counter=array-option is set, sparse should perform on par with vanilla Solr. The counter-updating step would be the first place to look for differences.
Measurements from UKWA suggests that sparse faceting on large result sets are markedly slower (x2) than vanilla Solr for some setups. The old option of switching directly to vanilla Solr when the result set exceeds a certain size should be re-introduced until the discrepancy between sparse & vanilla Solr has been eliminated.
Technical note: When the
facet.sparse.counter=array
-option is set, sparse should perform on par with vanilla Solr. The counter-updating step would be the first place to look for differences.