A skewed rows-per-value measure may not indicate a problem. An even distribution of rows across AMPs and evenly distributed workloads is possible when running a query against skewed data. With good statistics, the Optimizer can generate good query plans even when the distribution of table rows is skewed.
Data can be naturally skewed. Data that is not skewed is better, but Vantage is equipped to deal with skew, and can often process skewed data successfully.