I want learn about NoSQL and when is better to use SQL or NoSQL. I know that this question depends on the case, but I'm asking for a good documentation on NoSQL, and some explanation of when is better to use SQL or NoSQL (use cases, etc). Also, your opinions on NoSQL databases, and any recommendations for learning about this topic are welcome.
Topic nosql
Category Data Science
Check Martin Fowler's Personal website. He writes good and specially answers one of your questions in his book: "NoSQL Distilled"
Please have a look at my answer here:
Motivations for using relational database / ORM or document database / ODM
Short version:
Use NoSQL when data size and number of transactions per second forces it, which typically happens above a few tens of TB and millions of transactions per second (db in memory, running on cluster), or at hundreds of TB and thousands of transactions per second (traditional db on disk, transactions per second is highly dependent on the usage pattern). Traditional SQL scales up to that point just fine.
NoSQL is well suited for some problems (data has a natural sharding, schema is flexible, eventual consistency is ok). You can use for those even if scaling doesn't force you to.
Developer familiarity with tools and ops ease of deployment are major factors, don't overlook them. A solution may be technically better but you may have a hard time using it, make sure you need it and make sure you budget for the learning curve.
As to how to learn it: fire up a MongoDB image on AWS, or DynamoDB, and have fun!
Geeks Mental is a community that publishes articles and tutorials about Web, Android, Data Science, new techniques and Linux security.