MongoDB is an open-source database management system (DBMS) that uses a document-oriented database model which supports various forms of data. It is one of the numerous nonrelational database technologies which arose in the mid-2000s under the NoSQL banner for use in big data applications and other processing jobs involving data that doesn't fit well in a rigid relational model. Instead of using tables and rows as in relational databases, the MongoDB architecture is made up of collections and documents.

A record in MongoDB is a document, which is a data structure composed of field and value pairs. MongoDB documents are similar to JavaScript Object Notation objects but use a variant called Binary JSON (BSON) that accommodates more data types. The fields in documents are akin to the columns in a relational database, and the values they contain can be a variety of data types, including other documents, arrays, and arrays of documents, according to the MongoDB user manual.

Documents, which also must incorporate a primary key as a unique identifier, are the basic unit of data in MongoDB. Collections contain sets of documents and function as the equivalent of relational database tables. Collections can contain any type of data, but the restriction is the data in a collection cannot be spread across different databases.

The mongo shell is an interactive JavaScript interface to MongoDB which allows users to query and update data, and conduct administrative operations. The shell is a standard component of the open-source distributions of MongoDB. Once MongoDB is installed, users connect the mongo shell to their running MongoDB instances.

The BSON document storage and data interchange format used in MongoDB provides a binary representation of JSON-like documents. Automatic sharding is another key feature that enables data in a MongoDB collection to be distributed across multiple systems for horizontal scalability as data volumes and throughput requirements increase.

The NoSQL DBMS uses a single master architecture for data consistency, with secondary databases which maintain copies of the primary database. Operations are automatically replicated to those secondary databases for automatic failover.