NoSQL databases are not something recently introduced, it was another type of model that was discussed with relational database management system (introduced in 1970s) whenever disadvantages of relational database management concepts were heavily hitting against traditional database management systems. It is as old as other database concepts like file-based models and relational models and was exist since late 1960s.
While the relational database model is considered as the Second-Generation database model, NoSQL database model is considered as Third-Generation database model. Initially, NoSQL model was introduced for addressing Object-Oriented concepts within the databases calling it as Object-Oriented Database. This was not required with many solutions at the beginning, used by few applications but it started seeing as another type of required backend product with modern projects related to big data, real time applications, and databases that require heavy scaling out implementations.
What it exactly does?
There are different types of NoSQL databases; Column databases (Cassandra), Document databases (MongoDB, DocuemntDB), Key-value databases (Oracle NoSQL database), Graph databases (InfiniteGraph). NoSQL does not mean No Structured Query Language, it means no relational structure, hence No-Relational would be the best name for it. The type we talk about with this post is, Document databases, and it does not mean it stores documents, it stores objects, hence object database would be the best name for it.
Document databases store objects, with no predefined, static schema. It is flexible and designed for documents with semi-structures, typically schema-free JSON. One key advantage with NoSQL is, scaling out. Traditionally, relational database management systems support scaling up by adding more resources to the box, but limited with various factors like OS, Mother-board used, etc. Scaling out is also possible with relational databases but heavily limited. This is completely addressed with the NoSQL database. It is designed for scaling out and boundary for the limitation is availability of resources, nothing else.
How it stores data?
If you understand how it maintains data, then you understand NoSQL document database. We all are familiar with relational structure, we maintain relations (with tables), entities (with rows), and attributes (with columns) with our relational databases. NoSQL database maintains them differently; row as a document and column as a property.
Here is an example. Following image shows how a table is manged for Customers in a relational database.
And this is how they have be managed in NoSQL database.
As you see, though all three documents have common properties, addresses of customers are maintained different ways, this is because NoSQL is schema-free.
How Microsoft supports on it?
Microsoft implementation of Document database is called as Azure DocumentDB. If you have an Azure Account, you can create a DocumentDB and start using it for your NoSQL requirements. Since it is completely could-based, you do not need to worry on infrastructure required for it, setting it up and mataintaining it. Let's see the way of creating and using it with next NoSQL post.
Thanks It is really helpful. can you explain more about which kind of domain suitable for these various technologies. for an example if we decide to select db technology how we can determine which type of DB technology suitable for this domain etc.
ReplyDeleteRegards..
Hi Chanuri,
ReplyDeleteThis is all about whether we are handling structured data or not regardless of the domain. If we have some data to be processed, something like emails, a book, then NoSQL is better than relational. But data we have to work with has a defined structure, you can simply set columns for maintaining attributes, then relational is the best.
Regards
Dinesh