- Key-Value Store database
Temporary: as memcached. The temporary key-value database stores the data in memory, which in two cases results in the loss of data, the first is the power outage, but the data content exceeds the memory size. The benefits of this treatment are very fast.
Permanent type: such as Tokyo Tyrant, flare and Roma
The two are both types: Redis. Redis first data in memory, and then meet certain conditions (the default is 15 minutes more than 1 times, 5 minutes more than 10, 11 minutes more than 10,000 keys change), the data synchronization to the hard disk, so that the data processing speed, but also to ensure that the data is permanent. My doubts are: according to this particular condition, the data can still be lost, how is this situation solved? Or is there a scenario where the data is completely saved in use?
- Document-oriented database
such as MongoDB and COUCHDB. You do not define a table structure, and you can use complex query conditions to get data, but you do not support transaction processing and join operations.
- Column-oriented database
such as Cassandra, HBase, and hypertable belong to this type.
A column-oriented database is suitable for handling operations that read a few columns of a large number of rows and update the specific columns of all rows simultaneously. My question is: What kind of application scenario is it that should be rare?
NoSQL database Technology-The 1th introduction to NoSQL and Big data types of NoSQL databases
