October 2015 Summaries

MongoDB 3.2 introduces several aggregation enhancements, including the $lookup operator that implements left-outer equi-joins in the MongoDB Aggregation Framework. This allows developers to combine data from multiple collections within a single query, reducing complexity and improving performance. The Aggregation Framework is also being extended with new operators, such as array operations, standard deviations, square roots, logarithms, and more. These enhancements aim to extend the options for performing analytics on the live operational database, ensuring that answers are delivered quickly and reflect current data.

MongoDB has become an attractive choice for organizations due to its ability to handle modern, unstructured data growth at a rate of 40% annually, with 90% of that data being unstructured. This requires faster and more efficient application development. Several companies, including Telefonica and Shutterfly, have successfully migrated from traditional relational databases (RDBMS) like Oracle to MongoDB, achieving significant improvements in performance, scalability, and cost savings. For example, a small team at Telefonica was able to build a user data management system in just 3 months using MongoDB, while Shutterfly reduced its data storage costs by 80%.

The text discusses the operationalization of Apache Spark with MongoDB, a NoSQL database. The tutorial covers setting up a Spark environment with MongoDB, reading data from MongoDB using Spark DataFrames, and writing data to MongoDB. The benefits of using MongoDB as an input or output for Hadoop jobs are highlighted, including the ability to define secondary indexes for fast data retrieval and low latency reporting. The text also discusses how to run Spark queries on any slice of data in MongoDB without table scans, leveraging the power of Spark and the indexing capabilities of MongoDB. The tutorial demonstrates how easily the power of Spark can be combined with the power of MongoDB for operational requirements of analytics and a data lake environment.