The main change in this PR is that it adds code to "flatten" segments that have overflows - basically a forced reorg. We use this in the inserts-as-updates experiment to simulate having "perfect allocation" (we flatten the DB after running an insert workload). To reduce the amount of engineering needed, I did not implement the overflow flattening for when segments are disabled.
This PR also includes fixes to the CMake files and checks in the experiment definitions for the inserts-as-updates experiments.
The main change in this PR is that it adds code to "flatten" segments that have overflows - basically a forced reorg. We use this in the inserts-as-updates experiment to simulate having "perfect allocation" (we flatten the DB after running an insert workload). To reduce the amount of engineering needed, I did not implement the overflow flattening for when segments are disabled.
This PR also includes fixes to the CMake files and checks in the experiment definitions for the inserts-as-updates experiments.