1. MongoDB JSON Documents

MongoDB = JSON Database

Best for

✔ REST APIs

✔ User Profiles

✔ Product Catalog

✔ Dynamic schema

2. DynamoDB :

Cloud NoSQL Database by AWS.

Stores Key-Value Document

DynamoDB = AWS MongoDB (managed service)

Advantages

Very fast

Fully managed

Auto scaling

Serverless

Shopping carts

Sessions

3. Redis :

In-memory database.

Redis = RAM Database

User Session

OTP

Cache

Leaderboard

Shopping Cart

Millions of operations/sec

Data Types

String,List,Set,Hash,Sorted Set,Bitmap,Streams, TTL

Best for

Cache, Session, Distributed Lock, Rate Limiter, Message Queue

Hadoop vs HDFS vs HBase

Hadoop

├── HDFS (Storage)

├── HBase (Database)

├── Hive

├── Spark

├── MapReduce

4. Hadoop

Hadoop = Entire Big Data Ecosystem

Purpose

Big Data Framework.

It is NOT a database.

It is NOT storage.

It is a platform.

Contains

HDFS, YARN, MapReduce, Hive, Pig, Spark, ZooKeeper,etc.

5. HDFS

HDFS means Hadoop Distributed File System

HDFS = Linux File System for Hadoop

Purpose:    Distributed storage.

Stores :    Huge files.

Example :

10 TB CSV

100 TB Images

1 PB Logs

Features

Replication

Fault tolerant

Distributed

Cheap storage

Not suitable for

Random lookup Transactions

6. HBase

Purpose

Distributed NoSQL Database.

HBase = Oracle built on HDFS

Built on HDFS.

Stores

Billions of rows.

Random read/write.

CustomerID

OrderID

TradeID

Unlike HDFS

HBase allows

SELECT WHERE ID=100

HDFS cannot.

7. Data Lake

Data Lake = Warehouse for all company data

Store ALL raw data.

Everything goes into Data Lake.

PDF,CSV,Excel,Image,Video,Logs,Database Dumps,JSON,XML

Common storage

HDFS

Amazon S3

Azure Data Lake

Google Cloud Storage

8. Spring Batch

Batch Processing. Runs once.

Spring Batch = Offline Processing

Processes millions of records.

Example   Every night

Read 50 million trades

↓

Validate

↓

Transform

↓

Insert Oracle

↓

Generate Report

Not real time.

Usually scheduled.

    Wait

↓

Process Tonight

↓

   Finished

9. Event Streaming

Event Streaming = Real Time Processing

Purpose: Continuous data flow.

Unlike Batch Data comes Immediately.

Example: Credit Card

Everything happens instantly.

Receive Event

↓

Immediately Process

↓

Next Event

↓

Immediately Process

↓

Never Stops

Big Data Flow

ATM

↓

Kafka

↓

Spring Boot

↓

Redis Cache

↓

MongoDB

↓

Spring Batch

↓

HDFS

↓

HBase

↓

Data Lake

↓

Analytics

How They Work Together (Real Banking Example)

Imagine Standard Chartered Bank processing credit card transactions:

1. A customer swipes a card.

2. The transaction is published to Kafka (event streaming).

3. A Spring Boot microservice validates the payment.

4. Frequently accessed customer data is retrieved from Redis for low latency.

5. Customer and transaction records are stored in MongoDB (or another operational database).

6. At the end of the day, Spring Batch processes millions of transactions for settlement and regulatory reporting.

7. Historical transaction files are archived in HDFS.

8. High-volume historical transaction records requiring fast lookups are stored in HBase.

9. All raw files, logs, reports, and exports are retained in the Data Lake for analytics, compliance, and machine learning.

This end-to-end flow is a common architecture in large banking and financial systems