- Requirements and
features that are
included in an
SQL database system:
- Each relation defines
a very strick structure
Consequence:
- "Free format" data
cannot be stored
in a Relational Data model
(and can therefore not be processed by SQL)
- "Free format" data
cannot be stored
in a Relational Data model
- SQL enforces a
strick data consistency:
Consequence:
- Locking can
reduce the
level of
parallellism
- SQL cannot support millions of simultaneous users due to synchronozation bottlenecks
- Locking can
reduce the
level of
parallellism
- Each relation defines
a very strick structure
- Applications that
uses non-SQL like
data:
- Facebook:
- Data =
postings
that
do not have
a fixed structure
Can include images and videos
- Data =
postings
that
do not have
a fixed structure
- Facebook:
- Applications that
have millions of
simultaneous users:
- Facebook: (again)
- Fact:
- Many tech companies have faced
data management and
data processing
challenges for which
SQL is
not a
suitable solution
- Many ad hoc
(specialized)
data processing and
management solutions
have been
developed for
specific needs
- The collection of these specialized data processing and management solutions are now know as: Not Only SQL (NOSQL) systems
- Many tech companies have faced
data management and
data processing
challenges for which
SQL is
not a
suitable solution
- Examples:
- BigTable
- Google developed
BigTable to store the
vast amount of data in:
- Webpage indexing, Gmail, Google Map, etc...
- BigTable is a "column-based" data storage system
- Google developed
BigTable to store the
vast amount of data in:
- DynamoDB
- By Amazon
- Available in Amazon's
cloud services
- DynamoDB is a "key-value" data storage system
- By Amazon
- Cassandra
- By Facebook
- Cassandra is a hybrid system beased on: the key-value and the column-based system
- By Facebook
- MongDB
- Data storage system for documents in the JSON format
- NeoJ
- Data storage system for graphs
- BigTable