PostgreSQL has good enough built in search but I find it limited in large real world global user cases: - CJK support is tricky for almost all of these search

PostgreSQL has good enough built in search but I find it limited in large real world global user cases: - CJK support is tricky for almost all of these search products, but postgresql is probably the only one that needs additional extensions, and some extensions are not updated for a while. - Scaling is tricky, bc on one hand, your search and other normal queries may not have same load pattern, but you only have one schema in multiple nodes, so sometimes you have to waste some resource; on the other hand, dependencies of the 3rd party CJK extensions makes it hard to use managed instances and maintaining your own is time consuming. - Most other search engines have well tuned APIs for typical use cases like filters/typo/etc, but you have to build your own in PostgreSQL. Of course it’s not entirely bad thing bc PostgreSQL does have most flexible query capabilities for those less than typical use cases. End of day, search for end users have many semi-edge cases, one have to try all these engines to find the best fit. For large use cases, operational cost and user tolerance of search errors are all part of consideration. A conventional wisdom is do less work in the early days , and find better solutions later if have to.