Evaluation Issues

[zjs1224522500]

Firstly, thank you for sharing your great works : )

I tried to reproduce the evaluation results and I met some problems when evaluated Data Loading (Section 5.1.1 in paper).
To keep same configurations as the paper presents, I tried following commands:

# Generate same workloads 
./load_gen -I10000000 -E128 -L0.03125

# And I copied the workload.txt to the directory which contains `working_version`.

# Run with tiered compaction
./working_version -P512 -B128 -E128 -C2 > load_2.txt

# Run with full compaction
./working_version -P512 -B128 -E128 -C1 -c8 > load_1_8.txt

Questions：

1. In tiered compaction, the output results shows that #kv_in_tree is 3175417, not 10000000. Is it correct? Or I can ignore this results and only focus on the #I_done.
2. I am not sure that -C1 -c8 is right config cmd for Full Comaction, as I found that the bts_wr_cmpt (about 19GB) is lower than the original evaluation result in paper (about 43GB).
3. How to set config cmd for 1-Lvl compaction strategy?

Could you please give more details about evaluation processes for different compaction strategies? I'll appreciate it.

Thanks a lot~
Best Regards