blog

The Battle of the NoSQL Databases – Comparing MongoDB and Oracle NoSQL

Gayathri Mageswaran

Published March 23, 2020

Modern IT needs to have a non-relational, dynamic schema (meaning no requirements for Joins Statements Queries) to provide support for Big Data/real-time applications. NoSQL databases were created with the notion of improving data processing performance and tackling the ability to scale out to overcome distributed database load using the concept of multiple hosts have won the new generation demand for data processing.

Besides providing the essential support for various data models and scripting languages, MongoDB also allows easy to start with the process for the developers.

NoSQL database open the doors to…

Text-based protocols using a scripting language (REST and, JSON, BSON)
Indeed minimal cost to generate, store and transport data
Support huge amounts of data processing.
Increased write performance
Not required to perform object-relational mapping and normalization process
No rigid controls with referential integrity rules
Reducing maintenance cost with database administrators
Lowering expansion cost
Fast key-value access
Advancing the support for machine learning and intelligence

MongoDB Market Acceptance

The modern needs for Big Data Analytics and modern applications play a crucial role in the need to improve the lifecycle of data processing, with no expectations for hardware expansion and cost increase.

If you are planning for a new application, and you want to choose a database, arriving at the right decision with many database options in the market can be a complicated process.

The DB-engine popularity ranking shows that MongoDB stands at no.1 compared to Oracle NoSQL (which placed at No. 74). The trend, however, is indicating that something is changing. The need for many cost-effective expansions goes hand in hand with much simpler data modelling, and administration is transforming how developers would want to consider the best for their systems.

According to Datanyze market share information to-date there are about 289 websites that are running on Oracle Nosql with a market share of 11%, where else MongoDB has a complete 12,185 website with a market share of 4.66%. These impressive numbers indicate that there is a bright future for MongoDB.

NoSQL Data Modeling

Data modelling requires understanding of…

The types of your current data.
What are the types of data that you are expecting in the future?
How is your application gaining access to required data from the system?
How is your application going to fetch required data for processing?

The exciting thing for those who have always followed the Oracle way of creating schemas, then storing the data, MongoDB allows creating the collection along with the document. This means the creation of collections is not a must-have to exist before the document creation takes place, making MongoDB much appreciated for its flexibility.

In Oracle NoSQL, however, the table definition has to be created first, after which you can continue to create the rows.

The next cool thing is that MongoDB does not imply strict rules on schema and relations implementation, which gives you the freedom for continuous improvement of the system without fearing much on the need to ensure tight schematic design.

Let’s look at some of the comparisons between MongoDB and Oracle NoSQL.

Comparing NoSQL Concepts in MongoDB and Oracle

NoSQL Terminologies

MongoDB	Oracle NoSQL	Facts
Collection	Table / View	Collection / table act as the storage container; they are similar but not identical.
Document	Row	For MongoDB, data stored in a collection, in the form of documents and Fields. For Oracle NoSQL, a table is a collection of rows, where each row holds a data record. Each table row consists of key and data fields, which are defined when a table is created.
Field	Column
Index	Index	Both databases use an index to improve the speed of search carried out in the database.

Document Store and Key-Value Store

Oracle NoSQL provides a storage system that stores values indexed by a key; this concept is viewed as the least complex model as the datasets consist of an indexed key-value. The records organised using major keys and minor keys.

The major key can be viewed as the object pointer and the minor key as the fields in the record. Efficient Search for the data is made possible with the use of the key as the mechanism to access the data just like a Primary key.

MongoDB extends key-value pairs. Each document has a unique key, which serves the purpose to retrieve the document. Documents are known as dynamic schema, as the collections in a document do not need to have the same set of fields. A collection can have a common field with different types of data. These attributes lead the document data model to map directly to support the modern object-oriented languages.

MongoDB

Oracle NoSQL

Document store

Example:

Key-value store

Example:

BSON and JSON

Oracle NoSQL uses JSON as a standard data format to transmit (data + attribute-value pairs). On the other hand MongoDB uses BSON.

MongoDB	Oracle NoSQL
BSON	JSON
Binary JSON – binary data format – induces faster processing	Javascript Object Notation – standard format. Much slower processing compared to BSON.
Characteristics : BSON is not human-readable format Lightweight Traversable Efficient. Additional datatype: BinData and Date datatypes	Characteristics: Human-readable and writable format Lightweight Skim through all the content Text-based data interchange format Language independent.

BSON is not in a human-readable text, unlike JSON. BSON stands for binary-encoded serialization of JSON like data, mainly used for data storage and a transfer format with MongoDB. BSON data format consists of a list of ordered elements containing a field name (string), type, and value. As for the data types BSON supports, all the datatypes commonly found in JSON and includes two additional datatypes (Binary Data and Date). Binary data or known as BinData that is less than 16MB can be stored directly into MongoDB documents. BSON is said to be consuming more space than JSON data documents.

There are two reasons why MongoDB consumes more space as compared to Oracle NoSQL:

MongoDB achieved the objective of being able to traverse fast, enabling the option to traverse fast requires the BSON document to carry additional metadata (length of string and subobjects).
BSON design can encode and decode fast. For example, integers are stored as 32 (or 64) bit integers, to eliminate parsing to and from the text. This process uses more space than JSON for small integers but is much faster to parse.

Data Model Definition

MongoDB Collection Statement

Create a collection

db.createCollection("users")

Creating a collection with an automatic _id

db.users.insert

( {
    User_id: "U1",
    First_name: "Mary"                  
    Last_name : "Winslet",  

    Age       : 15

    Contact   : {

               Phone: "123-456-789"

               Email: "[email protected]"  

                }

   access  : {

              Level:5,

              Group:"dev"

             }            

})

MongoDB allows the related pieces of information in the same database record to be embedded. Data model Design

Oracle NoSQL Table Statement

Using SQL CLI to setup namespace:

Create namespace newns1;

Using namespace to associate tables and child table

news1:users

News1:users.access

Create Table with an IDENTITY using:

Create table newns1.user (

idValue INTEGER GENERATED ALWAYS AS IDENTITY (START WITH 1 INCREMENT BY 1 MAXVALUE 10000), 

User_id String,

First_name String,

Last_name String, 

Contact Record (Phone string,         

                Email string),

Primary key (idValue));

Create Table using SQL JSON:

Create table newns1.user (

idValue INTEGER GENERATED ALWAYS AS IDENTITY (START WITH 1 INCREMENT BY 1 MAXVALUE 10000),

User_profile JSON, 

Primary Key (shard(idValue),User_id));

Rows for User table: type JSON

{

  "id":U1,

  "User_profile" : {

     "First_name":"Mary",

     "Lastname":"Winslet",

     "Age":15,

     "Contact":{"Phone":"123-456-789",                   

     "Email":"[email protected]"

                   }

}

Based on the Data Definitions above, MongoDB allows different methods for schema creation. Collection can be defined explicitly or during the first insert of the data into the document. When creating a collection, you can define an objectid. Objectid is the primary key for MongoDB documents. Objectid is a 12-byte binary BSON type that contains 12 bytes generated by MongoDB drivers and the server using a default algorithm. MongoDB objectid is useful and serves the purpose to sort the document created in a specific collection.

Oracle NoSQL does have several ways to start defining tables. If you are using the Oracle SQL CLI by default, new table creation will be placed in sysdefault until you decide to create a new namespace to associate a set of new tables with it. The above example demonstrates the new namespace “ns1” created, and the user table is associated with the new namespace.

Besides identifying the primary key, Oracle NoSQL also uses the IDENTITY column to auto-increment a value each time you add a row. The IDENTITY value is auto-generated and must be an Integer, long or number data type. In Oracle NoSQL, IDENTITY associates with the Sequence Generator similar to the concept of objectid with MongoDB. As Oracle NoSQL allows the IDENTITY key to be used as the primary key. If you are considering IDENTITY key as the primary key, this is where careful consideration is required as it may have an impact over the insertion of data and the update process takes place.

MongoDB and Oracle NoSQL table/collection level definition show how the ‘contact’ information is embedded into the same single structure without requiring additional schema definition. The benefit of embedding a dataset is that no further queries would be necessary to retrieve the embedded dataset.

If you are looking to maintain your system in a simple form, MongoDB provides the best option to retain the data documents with less complication. At the same time, MongoDB provides the capabilities to deliver the existing complex data model from relational schema using schema validation tool.

Oracle NoSQL provides the capabilities to use SQL, like query language with DDL and DML, which requires much less effort for users who have some experience with the use of Relation Database systems.

MongoDB shell uses Javascript, and if you are not comfortable with the language or with the use of mongo shell, then the best fit for the process is to opt to use an IDE tool. The top 5 MongoDB IDEtools in 2020, like studio 3T, Robo 3T, NoSQLBooster, MongoDB Compass and Nucleon Database Master will be helpful to assist you in creating and managing complex queries with the use of aggregation features.

Performance and Availability

As the MongoDB data structure model uses documents and collections, using BSON data format for processing a huge amount of data becomes much faster compared to Oracle NoSQL. While some consider querying data with SQL is a more comfortable pathway for many users, the capacity becomes an issue. When we have a huge amount of data to support, the need for increased throughput and followed by the use of SQL to Design complex queries, these processes are asking us to relook at the server capacity and cost increase over time.

Both MongoDB and Oracle NoSQL provide sharding and replication features. Sharding is a process that allows the dataset and the overall processing load to be distributed across multiple physical partitions to increase processing (read/write) speed. The implementation of shard with oracle requires you to have prior information on how sharding keys work. The reason behind the pre-planning process is due to the need of having to implement the shard key at the schema initiation level.

The implementation of shard with MongoDB gives the room for you to work on your dataset first to identify the potential right shard key based on query patterns before implementation. As the sharding process includes data replication, MongoDB has a reputation for fast data replication as well. Replication takes care of fault tolerance of having to have all data in a single server.

Conclusion

What makes MongoDB preferred over Oracle NoSQL is that it is in binary format and its inborn characteristics of lightweight, traversable, and efficient. This allows you to support the advancing modern application in the area of machine learning and artificial intelligence.

MongoDB characteristics enable the developers to work much more confidently to build modern applications faster. The MongoDB data model allows the processing of huge amounts of unstructured data with an improved speed that is well thought off compared to Oracle NoSQL. Oracle NoSQL wins when it comes to tools it has to offer and possible options to create data models. However, it is essential to make sure developers and designers can learn and adapt to technology fast, which is not the case with Oracle NoSQL.