The idea of "adding a `tenantId` token to the index in some way, then apply a fulltext filter when running queries" you said in the issue description, technically can be done, but it practically it should not be used, since all tenant data will be store in the same index files, so the searching will include searching from ALL tenant data instead of related tenant needed. It will create a big performance impact on each search as the tenant/user base grows and each tenant data grows.
|