Verdict TODO

TODO List

In Progress

1. Supporting Spark SQL, Druid
2. Make Verdict's meta data location configurable
3. Let Verdict read configuration options from a file

Future

1. Suggesting samples after analyzing query logs
2. Supporting SSL (for non-Kerberos)
3. Making confidence interval estimates analytic
4. Supporting IN (subquery) correctly

Finished

1. Adding confidence intervals by Bootstrapping: I have pushed updates into dev repo; I will merge this into the main repository along with count-distinct feature.
2. Supporting COUNT(distinct *)
   1. Add universal samplers: By default, universe samples are built for top-ten high-cardinality columns.
   2. Currently working on using stratified samples for low-cardinality columns.
   3. If no appropriate samples are available, Verdict will use native HLL provided by Impala, Spark SQL, etc.
3. Supporting Kerberos
   1. Kerberos JDBC connection confirmed; will add this shortly.