Today, I am going to discuss the Graph Based Store Data Model of NoSQL, which is completely different than the previous three models.

Some of the most popular Graph-Based Databases are OrientDB, Neo4j, FlockDB.

What exactly it is?

There are main two words, which are the central part of Graph Based Store Database.

1. Node or Entity.
2. Edge or Relationship.

The main idea of this model is linking, and grouping related pieces of information.

This model represents the connections between two or more entities which are direct or indirect connected.

One of the best examples is:
The model of Social Network like Facebook Friend’s connections. In this model, directly or indirectly we can connect to N number of friends.
Another example of applications is Transport Links, Map and Spatial System, Routing System, Network Topologies System.

When we are trying to store a graph structure in RDBMS, We can store very limited and a single type of relation and adding an another relation type requires changing schema structure.

A Graph Based Store database is a schema-free and we can scale up to any level by adding a different type of Entities and Relations. (Nodes and Edges)

How exactly it works?

Each node is defined by a unique identifier and each edge represents a relationship between two nodes.

Each node can have an unlimited number of relationships, and each edge should be directional.

All the value is derived from the relationships, and it has not only a start-node and an end-node, but it has also set of properties like relation type, time of established, the distance between two nodes and other.

We can use these properties to query the graph.

Each node can have a different type of relationship which is defined by different ways like category, different time information, indexing, different path information.

It is very easy to adding a new relationship, changing existing nodes in the graph.

Now, today is a day to discuss one of the most important NoSQL Data Model that is known as Document Based Store Data Model.

Some of the most popular document databases are MongoDB, CouchDB, OrientDB, Terrastore.

A keyword “document” is a central word for Document Based Store Data Model.

It is very similar to Key-Value Store Data Model, and the difference is a whole document stored as a value which is referred by a unique key.

The Key always should be a unique and column name usually starting with “_id”.

We can store and retrieve the document using the different format like JSON, BSON, XML, and any other custom format.

It is also supporting query languages and the concepts of database index.

every document are self-describing and follow the hierarchical tree data structures.

We can also create a nested document like any other parent and child model. But unfortunately we cannot apply any JOINS in a query, and we have to manage this JOINS in an application side.

At any given point of time, we can add or remove columns from the document because this architecture is fully schema-free and flexible, in which each record can have different type of document data, and each document has an independent structure.

We can use this model for the real-time application like Blogging. In a blog, we have N number of articles which has N number of comments. Each article has multiple categories and author details.
For this kind of application, We require 100% scalability which is given by this model.

The sample document for Blog:

{

"_id": "Article_1",
"title": "What is NoSQL?",
"author": "Anvesh",
"category": "NoSQL",
"body": 
"A NoSQL database provides a mechanism for 
storage and retrieval of data which is modeled 
in means other than the tabular relations used 
in relational databases."
"comments":
   [
	["email": "neevan@gmail.com","body": "Nice post !"],
	["email": "roy@gmail.com","body": "Nice,I am sharing !"],
	["email": "martin@gmail.com","body": "The best post !"]
   ]

}

In this post, I am sharing a small note about the Key Value Store NoSQL Data Model.

The Redis and Riak are popular key-value store NoSQL Database.

What exactly it is?

In the RDBMS system, we have a table with unique key and values. We are also fetching values basis on a key.

The Key Value Store Data Model is also very similar to RDBMS structure.
In a Key Value Store Data Model, we have a key in left side and value in right side.

The Key Value Store Date model is same as a dictionary or hash model.

It is a schema-less format, and it is just about what you need to store.
A unique key value index the data, and each value can have independent format and stored as BLOB, JSON.

The group of keys and value stores into a Bucket and we can perform different manipulations on each Bucket.
The Bucket may have a different set of fields for every record.
Because of this structure, it is very flexible and scalable up to N level.

You can go with this:

• When we are doing a bulk insertion and write operations
• When we are creating simple and flat data model
• When we are reading the data using only Primary Key

You cannot go with this:

• When we are working with relationship related data
• When we are doing multikey operation like INSERT, UDPATE, DELETE using one KEY at a time
• When we are searching the data using a value

Now time to discuss about the different Data model of NoSQL Databases.

One of the most popular data model is, NoSQL Column Family Store Data Model.

Facebook’s Cassandra, Google’s BigTable Amazon DynamoDB and HBase are the most popular column stored base NoSQL Databases.

What exactly it is?

It allows you to store data with the key and mapped value to it, but these values are stored in Column Family.

A Column Family is a collection of rows, which can contain any number of columns for the each row.
You can create unlimited columns in a row; there are no any limitations.

The row key must be unique within a column family, but the same row key can be reused in another column family.

We can also use different data types for each row key.
e.g. User-1 wants to use a Mobile Number as row key, and User-2 wants to use an Email address as the row key.
This is possible.

Very simple term, we can say that Column Family is a logical division of similar data.
e.g., Column Family-1 for User’s Profile and Column Family-2 for User’s Friends.

A Column family is a very similar to the Table of RDBMS.

How exactly it works?

Keyspace: Name of the application. e.g.Gmail, Facebook.
ColumnFamily: Structure contained unlimited columns.
Column: Tuple with Name, Value, Timestamp.
Key: Name of the record.
SuperColumn: Containter of the more columns.

Each column has three attributes. The column name, column value and Timestamp. Timestamp automatically attached to the each column.

Using the Timestamp, we can also set expiring usage.
e.g. I want to delete a particular column in next 8000 second.
A real-time example is, Running a promo code or special offer for a set of users and running for the particular time. For this kind of application, we can expire a column by defining timestamp.

All the row keys and column stored in the particular order.

We can also create a Super column, which adds another level of nesting to the regular column family structure.

Each Column Family stores in a file and each file separately stores into Disk.
The benefit of storing data in the column is fast searching and accessing of the data.

For example, we have two column family one for User’s Profile and second for User’s Friends.
Now, I want to fetch User’s Profile data; then It accesses only a User’s Profile file which returns good performance.

Another good example is,
Cricketer’s award information. All cricketers have different award information like International, National and State level matches.
We can store an N number of different award information for each cricketer and stores vertically into Column Family.

This is my third day with the theory of NoSQL technology and understanding the core difference between Relational Database Management System and NoSQL technology.

The RDBMS products have been around us more than 25 years, and still, many enterprise applications are running on RDBMS products only.

But we should not avoid new terms like BigData, NoSQL which is the future of Data Storage and Data Science.

&nbsp

I marked some of the major differences between RDBMS and NoSQL technology.

The Relational database is entirely structured way of storing data, and the NoSQL is an unstructured way of storing the data.

The Relational database is fixed schema oriented and stores data into a tabular format. The NoSQL database does not depend on any other schema and stores data using a different architecture like key-value store, column base store, document base store, graph base store.

The Relational database uses Structured Query Language for manipulating the data and NoSQL does not follow any standard query language because it uses a different method of document reading and key-value pair matching.

The Relational database is vertically scalable, and NoSQL database is horizontally scalable.
We can add more hardware resources for vertical scaling of a Relational Database, and We can add more server or node for horizontal scaling of a NoSQL database.

The Relational database gives guarantee for every transaction by following the ACID Properties, and there are no any guarantees that NoSQL database is following ACID properties.

The Relational database is capable enough to handle complex query and relations of the tables. The NoSQL database does not govern by any standard so, in some situation, it cannot handle more complex queries.

The Relational database is capable enough to handle complex and nested transactions, but at this movement, NoSQL can process only simple transactions only.

We should not use NoSQL for the big ERP system because since 2009 lots of research is going on NoSQL technology and lots of improvement also requires for NoSQL technology.
We can prefer NoSQL technology for dynamic web application and mobile application like blogging, social networking, news board, Content Management System.

The compare to NoSQL databases, we can find big community support for Relational Database.

This is the second day, and sharing about why we should go with NoSQL?

I am using a computer and internet since last 15 years. I can easily compare past internet usage and current internet usage.

In the present situation, we are not simple internet users who are just using internet for browsing purposes.
Now we are the social internet user who is playing with tons of unstructured data like, Images, Videos, Comments, Tweets, and Social connections.

RDBMS products are suitable for specific application, but at a certain point, we should migrate with the BigData or NoSQL Technology.

Unlimited growth of Data:

In the real-time application, the volume of the data is increasing every day. A NoSQL technology is designed such a way that it can handle tons of data. It can handle different varieties of data which are coming very quickly from the different locations.

Performing well:

A NoSQL is performing well because it requires very fewer resources like Memory and CPU I/O. We can easily deploy a full version or partial version of the NoSQL database at any location so that nearest application can quickly access NoSQL database.

High Availability:

If we are experiencing with the RDBMS high availability concepts like Database Replication, Clustering and Mirroring, we can analyse this thing very easily because in some point of time we require acting like to make Database Offline.

NoSQL databases can provide continuous availability at a different location using the cloud. It doesn’t require to make your system or database offline for any modification or maintenance.

Scalability:

We can scale NoSQL database at any level and in any direction because there is not any fixed rule attached here.
Even If we divide the NoSQL database into different parts for different location and users, also it is working fine because it is not hard like any other database schema rule.

Flexibility:

A NoSQL is 100% flexible, it doesn’t require to depend on any schema rule. The data and storage also flexible because we do not require to follow any other rule like a Primary key and Foreign key. We can store any structured or unstructured data.
There are different NoSQL architecture available in the market, as per the situation we can use any type NoSQL database.

Today is my first day with NoSQL and sharing small note here.

Tons of digital data around us, and people are talking about the only Relational Database Management System.

In the past, it was a right discussion, but today this is a wrong discussion.

People may have wrong thinking that we have to use all the rule of Dr Edgar F. Codd’s for storing and manipulating the Data.

Tons of Non-Structured Data is around us because of social networking and eCommerce.( Data like picture, video, comments).
The RDBMS system is good for the structured data and worked with the fixed schema design.

What is NoSQL?

In simple term,
A NoSQL is a new way of thinking, and we can say NoSQL means storing and retrieving of data differently and simply design for horizontal scaling other than tabular relations format.

A NoSQL also referred as “non SQL”, “non Relational”, “not only SQL”.

A NoSQL is not a replacement of RDBMS product, but it is a good alternative for manipulating unstructured, unpredictable and messy data.

A NoSQL supports fast access to large binary objects like Video, Images, Docs and typically used for large database system where RDBMS product is not performing well to achieve a good query performance.

A NoSQL database is highly scalable because it does not follow any consistent rule or formula for storing and retrieving data. Even some of the NoSQL versions are violating the rule of ACID Properties.

In the market, there are different type of NoSQL databases available like Document based database, Graph-based database, Key-Value store database, Wide column based database.
As per the situation and requirement, we can choose and use the NoSQL database.

Some of the famous NoSQL are Google’s BigTable, Facebook Cassandra, Amazon’s Dynamo and SimpleDB and MongoDB.