Database Reference
In-Depth Information
Student Number
and
Class ID
in
Attendance
translate into foreign keys on the
Attend-
ance
RDBMS table because of the constraints from
Student
to
Attendance
and from
Class
to
Attendance
. This means that a
Student Number
cannot exist in
Attendance
without having this
Student Number
first exist in
Student
. A
Class ID
cannot exist in
At-
tendance
without having this
Class ID
exist in
Class
.
However, in MongoDB, if
Attendance
is a collection with a
Student Number
reference
to
Student
and a
Class
reference to
Class
, there are no checks to ensure the
Attendance
Student Numbers
and
Class IDs
point back to valid values in the
Student
and
Class
col-
lections.
SECONDARY KEY
Sometimes there is a need to retrieve data rapidly from a table to answer a business query
or meet a certain response time. A secondary key is one or more data elements (if there is
more than one data element, it is called a composite secondary key) that are accessed fre-
quently and need to be retrieved quickly. A secondary key is also known as a non-unique
index or inversion entry (IE for short). A secondary key does not have to be unique, stable,
nor always contain a value. For example, we can add an IE to
Student Last Name
in
Stu-
dent
to allow for quick retrieval whenever any queries require
Student Last Name
:
Student Last Name is not unique, as there can be two
Jones
; it is not stable and can change
over time; and sometimes we may not know someone's last name, so it can be empty.
MongoDB Non-Unique Index = RDBMS Secondary Key
The concept of a secondary key in relational databases is equivalent to the concept of a
non-unique index in MongoDB. Indexes are added to improve retrieval performance. There
are always tradeoffs with indexes, though. Balance retrieval speed with space (each index
requires at least 8 kilobytes) and volatility (if the underlying data changes, the index has to
change, too).
A query that does not use an index is called a “table scan” in an RDBMS and a “collection
scan” in MongoDB. This means that the server has to look through all of the data, starting
at the beginning, until all of the query results are found. This process is basically what
