Database Reference
In-Depth Information
These last two features—record types and repeated fields—distinguish
BigQuery from most relational databases, which can store only flat rows.
Records and repeated fields enable you to store the data in a more natural
way than you might in a relational database. For example, if your table
contains customer orders, you might want to store an entire order as a single
record, even though there were multiple items in the order. This makes it
easier to perform analysis of the orders without having to flatten the data or
normalize it into multiple tables.
Collections of rows of data following a single schema are organized into
tables
. These tables are similar to tables in a typical relational database
but have some restrictions. The only way to modify BigQuery tables is to
append to them or rewrite them—there is no way to update individual
rows. BigQuery also doesn't support table modification queries, like
ALTER
TABLE
,
DROP TABLE
, or
UPDATE TABLE
.
Collections of tables with similar access restrictions are organized into
datasets
. Many relational database systems allow you to have multiple
database catalogs. For instance, if you have a MySQL database with your
financial data you might want that to be a separate catalog from your user
data.
These catalogs map quite well to a BigQuery dataset. Datasets can be shared
with other users and groups. Collections of datasets owned by a single user
or organization are organized into
projects
. Projects are a Google Cloud
Platform concept that indicates a single billing entity with a team of users
in various roles. Chapter 4, “Understanding the BigQuery Object Model,”
discusses these abstractions in much more detail.
Distributed Cloud Computing
Google has a lot of hardware in their datacenters. A number of people
have tried to figure out just how many machines Google has by taking
into account things like global PC sales and maximum power capacity of
various known Google datacenters. It is a large number. Very few, if any,
organizations can match the scale and, as importantly, the organization
of Google's datacenters. Google's Cloud Platform allows people outside of
Google to take advantage of this scale and manageability.
