This paper describes the features of Sawzall, a new language that can
analyze data in parallel.

The strength of this paper is that it shows the basic syntax of this query
language precisely. Also the advantage of using Sawzall is introduced by
comparing with some other language or models, such as SQL and MapReduce.
Both the overall flow and the performance are discussed in the paper. Some
real examples together with corresponding programs are shown in the paper
to demonstrate that the language is really useful in practise.

Some weakness about this paper:
1. The motivation of inventing such new language is mentioned quite
late in the paper, till Section 13, so the readers are not quite clear
why this work is necessary, especially when user knows MapReduce in
advance.

2. To me, the advantage of this language comparing with MapReduce is not
significant enough. It seems that the benefit is that the program is
simpler to write by using this new language. But for a programmer who is
quite familiar with C++ and MapReduce, this benefit is not significant.

3. Some features, such as the ones mentioned in Section 11, are important,
but the readers are not convinced by the examples shown in this section.

4. Most examples are meant to solve simple problems. It will be better if
the authors can show that the language can also solve complicated problems
quite easily.

Generally speaking, the paper is worth reading, especially for persons who
want to know the advanced technology used in big search engine now. But
better motivation and more examples may be necessary for researchers who
needs to be convinced by its value.