In general we found the paper to present a good idea, with enough evidence that the technique is effective. Here are some outstanding points:
- The architecture were the system is going to be applied is not very well specified, we assumed it was a large scale SAN since it made more sense in terms of cost effectiveness.
- The writeback of the data from the surrogate RAID system maybe be buggy, since they delete the entry from the redirection table and then do the write, hence if this write fails data could be lost.