To start, I think these seem like reasonable milestones:
IMPALA-3705: Get existing Impala data loading working for Kudu tables
- Just the tables we absolutely need in the short term
- Run existing stress tests on Kudu tables
- Make changes to the existing stress test code
- Modify stress test workload to include some INSERT/UPDATE/DELETE queries. For now, let's just create some additional table (copy of something else) and just perform these on it randomly
Harrison and Michael should feel free to change this.