Original address: http://www.jdon.com/39240
The article summarizes the following points:
1. Write frequently, rarely read statistics, such as CTR, should use memory-based in-memory Key/value storage database such as Redis, or update-in-place text storage such as MongoDB.
2. Big data (such as weather data Business analysis data) can use
distributed database systems such as Hadoop.
3. Binary data (such as MP3s and PDFs documents), direct storage is sent directly to the client browser, such as Amazon S3.
4. Short-term data (such as Web sessions, lock status, short life cycle state) can be guaranteed to exist such as memcache. (Banq: In fact is in memory in-memory)
5. If you need to replicate data across multiple nodes (such as syncing music databases between Web apps and mobile devices), you can use replication features such as COUCHDB.
6. High-availability applications, high requirements for the machine time, even the zero machine. Automated clustering, redundant databases, such as Casandra and Riak.
When should I use a NoSQL storage database?
