Data Management in Cloud Computing: Storage, Replication, and CAP Theorem
What is Cloud Data Management?
In traditional systems, all company data is stored on one big hard disk inside a single database server in the office room. But in the cloud, millions of users read and write data simultaneously from all over the world.
Cloud Data Management is the system of organizing, storing, backing up, securing, and synchronizing massive amounts of structured and unstructured data across distributed data centers across the globe.
1. Key Challenges of Cloud Data
- Massive Scale (Big Data): Billions of gigabytes generated every day by apps like Instagram, YouTube, and UPI.
- High Availability: Data must be readable even if one entire data center catches fire or loses internet.
- Data Consistency: If you transfer ₹500 via Google Pay, your balance must update immediately on all banking servers so you cannot spend the same ₹500 twice.
2. Data Replication (Why Copy Data?)
Replication means creating identical copies of your data and storing them on multiple servers or in different geographical locations.
Benefits of Replication:
- Fault Tolerance: If Server A crashes, Server B immediately handles user requests without anyone noticing.
- Low Latency (Fast Speed): A user in India downloads files from a Mumbai data center copy, while a user in London downloads from a Frankfurt copy.
3. The Famous CAP Theorem (Must for Exams!)
Proposed by computer scientist Eric Brewer, the CAP Theorem states that a distributed data system can provide at most TWO out of the following THREE guarantees at the same time:
In real-world cloud systems, network cables and routers will occasionally fail, so Partition Tolerance (P) is mandatory. This leaves providers with a choice:
- CP Systems (Consistency + Partition Tolerance): If the network breaks, the system rejects requests rather than showing old data (e.g., Banking apps, ATM transactions).
- AP Systems (Availability + Partition Tolerance): The system stays online and serves data, but some users might see slightly delayed updates for a few seconds (e.g., Twitter/X likes count, YouTube view counts).
4. ACID vs. BASE Model
| Feature | ACID (Traditional RDBMS like MySQL, Oracle) | BASE (Cloud NoSQL like MongoDB, DynamoDB) |
|---|---|---|
| Meaning | Atomicity, Consistency, Isolation, Durability | Basically Available, Soft state, Eventual consistency |
| Consistency | Strict immediate consistency | Eventual consistency (data becomes consistent after a short delay) |
| Scalability | Vertical (buy bigger, more expensive servers) | Horizontal (add hundreds of cheap cloud servers) |
| Best For | Banking, finance, flight bookings | Social media feeds, product reviews, chat history |
💡 DAVV Exam Tip
Question (10 Marks): “Explain the CAP Theorem and discuss how data management is handled in cloud systems.”
- Clearly state: “You can only choose 2 out of 3: Consistency, Availability, Partition Tolerance.”
- Explain why Partition Tolerance (P) is compulsory in cloud networks.
- Compare ACID vs. BASE properties using the table above.