Chapter 5System Design~1 min read

Database Design आणि Scaling

SQL vs NoSQL, Sharding, Replication

Database हे system design मधला सर्वात critical component आहे. Wrong database choice, poor schema design — या चुका नंतर fix करणे खूप कठीण असतं. SQL vs NoSQL, vertical vs horizontal DB scaling — हे decisions मोठ्या systems मध्ये फार महत्त्वाचे असतात.

SQL vs NoSQL — कधी काय?

Comparison

text

SQL (PostgreSQL, MySQL):
✅ ACID transactions — banking, e-commerce साठी
✅ Complex JOINs, relationships
✅ Strong schema — data integrity
✅ Mature, battle-tested
❌ Horizontal scaling कठीण
❌ Schema changes painful
Use when: structured data, transactions, complex queries

NoSQL (MongoDB, Cassandra, DynamoDB):
✅ Horizontal scaling सोपं
✅ Flexible schema — schema-less
✅ High write throughput
✅ Specific use cases मध्ये खूप fast
❌ JOINs नाहीत (denormalization)
❌ ACID guarantees कमी (mostly)
Use when: unstructured data, massive scale, specific access patterns

Database Replication

Replication म्हणजे एका database चे data दुसऱ्या DB वर copy करणे. Primary-Replica setup: Primary वर writes होतात, Replicas वर reads. यामुळे read load distribute होतो आणि failover होते.

Primary-Replica Replication

text

Write Operations → [Primary DB]
                         ↓ (replication)
              [Replica 1]  [Replica 2]  [Replica 3]
                  ↑             ↑             ↑
              Read           Read           Read

Benefits:
- Read requests Replicas वर जातात — Primary वरचा load कमी
- Primary fail झाल्यास Replica promote होतो
- Geographical replication — different regions मध्ये copies

Database Sharding

Sharding म्हणजे data horizontally partition करणे — एकाच DB मध्ये सगळं न ठेवता multiple DB instances मध्ये वाटणे. User_id 1-1M → Shard 1, 1M-2M → Shard 2.

Sharding Example

text

Users Table — Too large for one DB (1 Billion rows)

Shard Key: user_id
Shard 1: user_id 0 - 250M
Shard 2: user_id 250M - 500M
Shard 3: user_id 500M - 750M
Shard 4: user_id 750M - 1B

Challenges:
❌ Cross-shard JOINs — खूप complex
❌ Resharding — shard full झाल्यावर rebalancing
❌ Application-level routing लागतं

💡

Database scaling order: 1) Cache वापरा (Redis), 2) Read Replicas add करा, 3) Vertical scaling, 4) Sharding — as a last resort. Sharding खूप complexity आणतो.

✅ Key Points — लक्षात ठेवा

▸SQL: ACID, joins, structured data — Postgres/MySQL
▸NoSQL: scale, flexibility — MongoDB/Cassandra
▸Replication: Primary writes, Replicas reads
▸Sharding: data horizontally partition करा
▸Cache first, scale DB later

0/11 chapters पूर्ण

मागे — Previous

Caching — Speed Up करा

पुढे — Next

Message Queues आणि Async Processing