NoSQL Databases - Recap

NoSQL vs. SQL

We began this lesson by comparing and contrasting Relational Databases/SQL with the various NoSQL database options that are out there, and outlined examples for when we would want to use each.

The 4 different kinds of NoSQL Databases are:

• Document Stores
• Key-Value Stores
• Column Stores
• Graph Databases

RDDs and Hadoop/Spark

We also took some time to dig into another data storage paradigm, Resilient Distributed Datasets, or RDDs such as Spark and Hadoop. We explored how these databases use redundancy to ensure that data is Fault Tolerant, and examined how the driving idea behind these paradigms , MapReduce, gives us great speed ups when working with truly massive datasets in a distributed manner.

MongoDB

Next, we explored how to install MongoDB and get the MongoDB Server up and running. We learned some commands that we could use to interact with mongo right there in the server shell, and then focused on connecting to our MongoDB server with Python.

You should remember the following high-level takeaways from MongoDB:

• You use the pymongo library and a little boilerplate to connect to the MongoDB database.
• In MongoDB, records are referred to as Documents.
• We can create groups of documents called Collections.
• Records are inserted by passing in our data as Python dictionaries.
  • We can insert a single record with our collection object's .insert_one() method.
  • We can insert many records by creating a list of dictionaries and passing the entire list to the collection object's .insert_many() method.
• We can write queries by using the collection object's .find() method.
• We can filter our queries by passing in the key-value pairs to filter by.
• We can also use logic on our filters by making use of modifiers.