These very fast databases, all different, do not organize along the “counting entities” (rows in a table) model. Instead, they store documents of various sorts in collections of loosely related documents, in turn stored in databases with loosely related collections.
The five top players (all open-source except DynamoDB):
- MongoDB
- Cassandra
- HBase
- DynamoDB (Amazon product offered in their AWS)
- CouchBase
Why No SQL? Actually, storage and retrieval of data is done by object-oriented methods that replace SQL, so that isn’t really the point. The real point is that these databases store and retrieve massive amounts of unstructured data. The developer can literally make up a new JSON or document type on the spot, in minutes, and not have to spend hours carefully making a highly structured RDBMS (and the surly DBA watching over it) happy.
This is the fast and sloppy way to do business, and it works just fine for what it does. Good luck applying structure to such a thing after it grows large! MDM probably won’t play nice with NoSQL.
No comments:
Post a Comment