We are interested in using the experimental new index version tsi1 available in the 1.3 line. It sounds like there is some level of interoperability with old data (such that index files begin to be created for data going forward but one can still query the old), so I am curious if the index can be enabled selectively by database/rp? We only need high cardinality for select things and it would be good to not use experimental features for the entire influxdb system. I suppose we could run another influxdb alongside, but ideally we don’t need to.
If there is not support for this, is there any plan to allow this type of configuration in the future?