Hacker News Reader

At least for our use case Postgres full text search wasn't good enough. Indexing and searching 1M records was troublesome, and its advanced queries weren't as flexible as we'd like, so after a while we started using Solr, and while it was better it still had its share of problems, so we are currently using ElasticSearch.

View on HN

I've gone down the same route, from PostgreSQL search to ElasticSearch with usually pretty good results. A friend has strongly recommended Algolia, so I'll explore that for the next project that needs it.

Hmm, was it "troublesome" in terms of performance building the index or doing the search or was there some other issue?

Part of the problem was due to the Django package we used to interface with the Postgres search, it wasn't designed to handle large tables. I don't recall how long the index took to build (hours? a day?), but the queries themselves were slow due to their complexity and ranking using ts_rank.

Interesting, thanks for the info. What did you switch to? Were the queries significantly faster?

Reply on news.ycombinator.com