Hi Bram, All,

If you really need fast CONTAINING for really large datasets,
so that some complications to achieve this are OK, you
can borrow a technique used (among others) in DNA databases.

Generated auxiliary table linking (uppercased, noaccented)
(indexed) triples of adjacent characters to the primary key
of your table.

Then, in your example query for candidates having both "BRA"
and "RAM", and depending on what you want to retrieve maybe
also " BR" and "AM ".

The intersection should be small enough for quick scanning
of true matches.

Depending on your data you even may start with Quadruples.

Of course this only makes sense if you have a lot of such queries.

Regards,
Peter Jacobi