An overview of NoSQL databases /
Joe Celko.
- [2016]
- 1 online resource (1 streaming video file (2 hr., 35 min., 12 sec.)) : digital, sound, color
Title from title screen (viewed December 1, 2016). Imprint from resource description page.
Presenter, Joe Celko.
"The traditional SQL database evolved from file systems. It assumes that all data can be expressed in scalar values in the columns of the rows of a table, just as file systems used fields in records of files. But the truth is that not all data fits into a traditional SQL model. The storage models have changed. Columnar databases have no rows, but keep tables as columns which can be assembled into rows if needed. This means that query optimizations are very different. The MapReduce Model and Key-Value Stores do not have tables at all. Cloud storage is not like traditional, local file systems. Parallel processing and brute force replace clever indexing optimization. Traditional SQL assumes that the data is static; think about common business applications like accounting and transaction processing. Changes to the data are explicit actions done by the database. Actions against the data are taken in other tiers of the application. But Streaming Databases (also known as Complex Event models) assume that data is constantly changing on its own. This would include sensor monitoring, stocks, commodity trades, and so forth. These events occur so fast that responses have to be built into the database. Graph Databases do not have a table concept at all! They model pure relationships, expressed as a graph structure. SQL assumes that we know the relationships in the data. Graph databases discover the relationships. They are used for social network analysis, patterns in various kinds of flows. Think about a SQL query to find the minimal set of 'the cool kids' in a social network whose adoption of a product makes it into the next trend."--Resource description page.
CL0500000804 Safari Books Online
Non-relational databases. SQL (Computer program language)