article was written by Nitin Agrawal and Ashish Vulimiri. The authors talk
about SummaryStore. They create this approximate time – series store to manipulate
large amount of time – series data on only one node and to create analytic
results through some input. In addition, they mention the reasons of why they
use a Temporal Analytics store to achieve their goals. Also, they describe the
architecture of this method. They emphasized that SummaryStore is created for analytics
and machine learning jobs. Moreover, they present some graphs and tables to
show the usage of the SummaryStore.
main problem that the authors try to show to the readers is that there wasn’t
any storage system such as SummaryStore before. At the past, the storage systems
which were used to manipulate such large volumes of time – series were
crashing. Moreover, another one problem that I think was important for authors
to pay attention is that they must show their job using real inputs like forecasting
and they did this and I think that’s why they achieve their goals.
contributions that I recognized while I was reading this article are many. Firstly,
SummaryStore helps in time – decayed. This time – decayed is a new abstraction
of summarizing data. It does this by using an algorithm to make more effective
the range of queries. Moreover, it shows some errors on its job and gives some
answers using the methods of machine learning. Some more contributions that I realized
reading the article are that they create a mechanism to integrate data for
answering on some queries that they want to have an answer. At least, another
one contribution is that when a question appears for the SummaryStore, it uses some
techniques to create the best answer for this question, which are not used
article was a good written one with sections for each theme they wanted to describe.
I think that using the figures, equations and graphs was a very good decision
because they let us know more specific how they worked to achieve. The only drawback
I recognized reading this article was at Figure 9. I think there was a lot of
information in one page and I get confused trying to realize what they want to
know from this page. But in general it was a very good article with a lot of knowledge
to get off.