The source code of the proposed algorithms as well as the code for the workload generation was included in the package that the authors provided. Documentation of how to run the code and rerun the experiments was clear and written in a step by step fashion. Although there wre some difficulties in running the code in the beginning, the authors provided helpful responses and I go tot run the code easily.

The plot creation scripts were also provided with the code and plotting after running the experiments was pretty easy as running a single command for each figure was enough.

The successfully repeated experiments fully reflect the results presented by the authors in their paper. The only difference is absolute the number of transactions performed which changes probably because my cluster is faster than that of the authors. However, I was unable to reproduce Figure 8 (TPC-C Throughput Varying Warehouse) as the system did not work for my infrastructure - it stalled after a number of experiments and the authors were not able to reproduce and debug it. As everything else was reproducible, I consider the non-reproducible figure not to be of high importance compared to the quality of code and scripts as well as the rest of results that were reproduced. 

As far as workability is concerned, the authors provided information on how to change the number of replicas and machines and run again the system. I have reproduced figure 4 with 4 machines instead of 2 (and with 2 replicas instead of 4). Figure 4 is not really affected by the different setup as the tendencies remain the same.

I have attached all the reproduced figures into the discussion.

Concluding, I consider the results completely "reproducible" and "workable".