Developer.com covers the buzz of NoSQL highlighting Graph Databases and Map/Reduce solutions that support big data and high scaling applications.
There are multiple types of NoSQL databases but they mainly fall into two camps; Key/Value stores and Graph databases. This is an oversimplification but it is a good starting point for further exploration. A key/value database stores documents, objects or simple values looked up primarily using a single key called a Document ID in most databases of this type. Most of these types of databases utilize Map/Reduce to produce a view, which is functionally similar to a SQL index. There are variations on this theme including document, multi-value, tuple and more but these type of stores share the common thread that they are primarily optimized to allow for the efficient retrieval object-like structure by a single key. Graph databases store data in terms of nodes and relationships. Graph databases require an entry point that is found through some sort of strong key relationship or index and is then explored relative to that entry point. Graph databases excel at friend-of-a-friend (FOAF) queries and data that is best described in terms of how it relates to other pieces of data such as social networking. One of the core philosophies of NoSQL in general is to use the right kind of database for the right kind of problem. There is no one SQL/NoSQL database that can both support real time applications scaling to millions of concurrent users while providing real time shopping cart analysis of that same user base. You will need to devote some time to understanding each kind of SQL and NoSQL database and think about the kinds of problems your business needs to address in order to find the right mix of tools to solve your business needs.
Keywords: Map Reduce • MSSQL