Impedance Mismatch problem: O/R issues
SQL: does not work well with large clusters
Google: Bigtable
Amazon: Dynamo
=> NoSQL
Origin: a Twitter hashtag to advertise a single meeting #nosql
- non-relational
- open-source
- cluster-friendly
- schema-less
- when you are getting a price from a row, you are utilizing schema, but when you don't have schema there is no guarantee data is there
- 21st Century Web
Data Model:
- 1 Aggregate-oriented
- 1 Document (usually JSON, you could do it in XML as well)
- there is an id, simillar to key in key value
- 1 Key-Value
- you can have metadata, and you can index them
- 1 Column-Family
- row key, column key
different data on different cluster based on the aggregate
aggregate = key
aggregate = document
changing aggregate is pain if you want to change the data and aggregate
- Graph