Statistics-based query optimisation #269
Labels
Comments
kal
pushed a commit
that referenced
this issue
Jan 20, 2016
This work is still incomplete though as testing shows that the optimizer isn't really doing a good job. This is primarily (I think) because we don't have enough stats. See issue #269
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Related to #111
This is a note to myself for future reference...
Tried using the existing store stats for optimization purposes, but currently we aren't really collecting enough statistics. We currently only have a total statements count for each predicate. Ideally we should also collection the number of distinct subjects and objects for each predicate as that would allow us to do a rough estimate of the weight of a pattern with any two of s, p, and o bound. With the current statistics we have to use default weightings for s and o which can throw off the optimizer quite badly.
It would be worth looking at updating collection of store statistics and then revisiting this.
The text was updated successfully, but these errors were encountered: