We need a generic benchmark tool to automatically execute a query set on a given data set with a given table schemas.
The requirements are as follows:
- A user can specific the benchmark (or test) set, which is defined as a schema set, a input data set, a query set, and a result set (optional).
- For each query, query time, the result validation, plan and others (user defined measurements) can be recorded.
- some options to handle query failures should be supported.
- e.g., abort if any failure or skip failed query
- A user can set the Tajo cluster to execute the benchmark (or test) set.