MicrosoftSome weeks ago on a tech call you shared with us a script to compute the skew for tables. I notice that the focus is on CCI tables, but wonder if the same should be applied to CI tables as well. Here is the scenario:
When focusing on a CCI table that is skewed, I notice that the table should not even be a CCI as it has less than 2 million rows, but it should remain distributed as it has more than 2GB. I therefore thought I would convert it to a CI table instead and leave the distribution as is.. However it occurred to me that even thought the table is not longer a CCI it will still be skewed as it is distributed using the same hash key.
So the question perhaps should be:
Does the index type matter when considering skew?
Thanks,
Anthony
