Database Reference
In-Depth Information
Chapter 2
BigQuery Fundamentals
This chapter introduces you to BigQuery, describing what it can do, when to
use it, when
not
to use it, and even a bit about why it is so fast. Because you
may be evaluating whether BigQuery is the right tool for you, this chapter
spends a bit of time comparing it to other systems and other architectures.
It discusses the performance and types of workloads best-suited to BigQuery,
and also how BigQuery fits with other offerings in the Google Cloud Platform.
This chapter concludes with an introduction to an AppEngine and Android
App that is used as the basis of many of the examples throughout the topic.
This sample app demonstrates several ways that BigQuery can be integrated
into an application—from log collection and analyses to dashboard
development and correlation of multiple data streams.
What Is BigQuery?
BigQuery, like many tools, started with a problem. Google engineers were
having a hard time keeping up with the growth of their data. The number of
Gmail users is in the hundreds of millions; by 2012, there were more than
100 billion Google searches done every month. Trying to make sense of all
this data was a time-consuming and frustrating experience.
Google is hugely a data-driven company. Decisions ranging from café menus
to interview strategies to marketing campaigns are made by analyzing data.
If you have a great idea but you don't have data to back it up, you're going to
have a hard time convincing anyone to implement your suggestion. However,
if you have data on your side that says people click more ads with a particular
shade of blue background, your shade of blue will likely become the new
official standard.
As Google grew exponentially, the amount of data available also grew
exponentially. Despite spending a lot of money on hardware and software for
relational databases, it was often difficult to ask simple questions of the data.
Despite having invented MapReduce to help analyze large datasets, it was
still difficult to get answers interactively, without waiting minutes or hours
for a long batch job to complete.
