Gian, Does the index use ordinary unicode collation (so that ‘d’ < ‘e’ … ‘z’ < ‘A’ < ‘D’, and therefore ‘druid’ < ‘dRuid’ < ‘fluid’ < ‘Druid’ < ‘DRUID’)?
There’s another collation where ‘d’ < ‘D’ < ‘e’ < ‘E’ etc. This puts ‘druid’, ‘dRuid’, ‘Druid’, ‘DRUID’ next to each other, so that the query "UPPER(name) = ‘DRUID’” can be answered with a very small range scan. Julian > On Apr 11, 2025, at 9:24 AM, Gian Merlino <g...@apache.org> wrote: > > String indexes in Druid are case-sensitive. UPPER (and LOWER) can use the > index, but they must do it through index scans, rather than point index > lookups. So index usage will be less efficient. > > Gian > > On 2025/04/09 18:50:45 vikas thakur wrote: >> On Wed, Apr 9, 2025 at 2:49 PM vikas thakur <vikas.thaku...@gmail.com> >> wrote: >> >>> Can someone help with logic of upper sql function of Druid ? What is >>> underlying implementation, does it use case insensitive indexes ? >>> Currently, benchmarking performance with upper function and standardizing >>> the casing for APIs. >>> >>> Regards, >>> VT >>> >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org > For additional commands, e-mail: dev-h...@druid.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@druid.apache.org For additional commands, e-mail: dev-h...@druid.apache.org
