Faceting over even small result sets and producing the relevant counts is really hard for postgres, and caching tends to be impossible due to the combinatorics

Faceting over even small result sets and producing the relevant counts is really hard for postgres, and caching tends to be impossible due to the combinatorics of the problem. There's some discussion around supporting better indexing options here that is relevant: https://www.postgresql.org/message-id/CAJqPDh9o%3DoxRP%3DxZhbCzzAs4Vr5hNUkYLMspGee%2B%3DmgbqG3tAg%40mail.gmail.com https://www.postgresql.org/message-id/CAJqPDh9o%3DoxRP%3DxZh.... Practically, https://github.com/cybertec-postgresql/pgfaceting https://github.com/cybertec-postgresql/pgfaceting is the best bet for people who must use postgres even though it still uses a manual maintenance step.