I agree with your complain regarding the cost of frequency partitioning not been explicitly mentioned in
the paper. Also the results are presented in a confusing manner. Regarding the confusion in phrases I
would like to know more the specifics of what you mean.
I liked the way you summarized the paper, it would be nice if you also stated separately in your opinion
the main contribution of the paper was. I fully agree with your point that having a prototype in memory
doesn’t make a strong case if the technique is being advertised for DSS applications. Regarding query
concurrency the authors do not provide any discussion as well.
I agree with most of your complains about the paper. The authors indeed ignore providing any
information regarding query updates and repartitioning, Even though DSS are not so much write
intensive the workload routinely requires appends. Also the paper was not well written in general. In my
opinion comparing against CJoin would not be a fair comparison as the focus there was improvement of
query concurrency, while Blink claims to focus on query predictability. In general I like your review, but
the summary was on the longer side and could be compressed.
I liked how you summarized the main points of the paper, if you added the main contribution section as
mentioned in the review guidelines it would make it would improve it further. I totally agree with all of
your complains regarding the paper.
Regarding complains about the paper not being well written and failure to discuss the start up cost their
system are justified. I ‘m not sure what you exactly mean by complicated queries in your third
complaint. Overall your review is well written, but lacks the main contribution section directed in the
review guidelines.
The reviewer provides a very clear and brief summary of the paper. I agree with the complains of the
reviewer regarding the authors not being able to motivate the use case of constant time queries
effectively. However regarding the second complain I would like to point out that the authors were not
advertising web searches as their “killer” applications, but rather wanted to explain the idea of constant
time query processing by this example.
I liked the first point regarding the strengths of the paper, I agree even if the paper has a lot of short
comings the main contribution of the paper is about how they emphasized the concept of uniform query
response time. I don’t agree with the second complain regarding databases being built mainly on
indexes , for DSS applications its mostly not the case and sequential scans are preferred. Overall the
review is well written and to the point