How do you choose between SQL and NoSQL databases?

Choose SQL (PostgreSQL, MySQL) for structured data requiring ACID transactions, complex JOINs, and strict schema enforcement. Choose NoSQL (MongoDB, DynamoDB, Cassandra) for flexible schemas, massive horizontal scale, high write throughput, and eventual consistency workloads.

When should you choose Vertical Scaling vs Horizontal Scaling?

Start with vertical scaling (adding CPU/RAM to one server) for simplicity. Transition to horizontal scaling (adding more servers) when hardware limits are reached, cost curves become prohibitive, or high availability is required. Horizontal scaling introduces distributed system complexity.

What is the difference between Sharding, Partitioning, and Indexing?

Indexing speeds up read queries within a single table using B-Tree or Hash structures. Partitioning splits a single table into logical physical chunks within one database instance. Sharding distributes rows across multiple distinct database servers using a shard key, scaling write throughput and storage.

How do you scale WebSockets to millions of concurrent connections?

Increase Linux file descriptor limits, use Layer 4 load balancing with sticky sessions, and implement a Redis Pub/Sub backplane so any WebSocket server can relay messages to clients connected to other servers. Pattern: Client → HAProxy → WS Servers ↔ Redis Pub/Sub.

What is the difference between LSI and GSI in DynamoDB?

A Local Secondary Index (LSI) shares the same partition key as the base table but uses a different sort key. It is limited to 10GB per item collection and supports strong consistency. A Global Secondary Index (GSI) uses a completely different partition key, spans all partitions, supports only eventual consistency, and has its own dedicated throughput capacity.

Which database should I use for transactions, user profiles, logs, and analytics?

Use PostgreSQL for financial transactions (ACID compliance). Use MongoDB for user profiles (flexible JSON). Use Cassandra for logs and high-throughput writes (LSM tree). Use Apache Druid for real-time analytics (columnar storage). Use Elasticsearch for search and catalog (inverted index). Use Redis for sessions and caching (in-memory).

How does Consistent Hashing work?

Consistent hashing maps both servers and keys onto a circular ring. A key routes clockwise to the nearest server. When a node is added or removed, only 1/N keys need to migrate (vs nearly 100% in modulo hashing). Virtual nodes (100-200 per server) ensure even distribution and prevent hotspots.

How do you design a distributed rate limiter?

Use a Token Bucket algorithm for burst-friendly limiting. For distributed enforcement across multiple API gateway instances, use Redis with Lua scripts to atomically evaluate and increment counters, preventing race conditions (dirty reads) where concurrent requests could bypass the limit.

How do you design a real-time distributed leader election mechanism using Apache ZooKeeper or etcd? — System Design Interview Answer & Trade-offs

Architectural Breakdown & AnalysisFAANG Aligned

In distributed microservice clusters with background workers (e.g., scheduled cron runners, payment reconciliation workers, partition managers), running multiple concurrent active instances causes race conditions and data corruption. An automated Leader Election mechanism guarantees that exactly one active master executes work, with instant failover if the master crashes.\n\n### 1. Leader Election via Apache ZooKeeper (Ephemeral Sequential Nodes)\nZooKeeper maintains a hierarchical in-memory tree of nodes called znodes:\n\n1. Create Sequential Nodes: All candidate worker processes attempt to create an Ephemeral Sequential znode under a common parent path /election:\n /election/node-0000000001, /election/node-0000000002, /election/node-0000000003...\n2. Elect the Smallest Sequence: Each candidate queries children of /election. The node with the lowest sequence number is elected Leader.\n3. Avoid the Herd Effect (Watch Predecessor Only):\n * If node 2 is not the leader, it does *not* watch the leader node 1!\n * Instead, node 2 sets a Watch on its immediate predecessor: node 1.\n * Node 3 sets a Watch on node 2.\n * If the leader (node 1) dies, its ephemeral znode vanishes. Only node 2 is notified, promoting node 2 to leader without waking up all other hundred candidate nodes.\n\n### 2. Leader Election via etcd (Leases & Revision Watches)\netcd simplifies this using TTL Leases and atomic Key revisions:\n1. Grant Lease: Worker acquires an etcd lease with a 5-second TTL: etcdctl lease grant 5.\n2. Atomic Put with Lease: Worker attempts to create /services/leader attached to the lease using a transaction (Txn).\n3. Success: If successful, worker is Leader and continuously sends KeepAlive heartbeats to maintain the lease.\n4. Failover: If the leader crashes, heartbeats stop. After 5 seconds, etcd revokes the lease and deletes the key, triggering a Watch event on standby workers to elect a new leader.\n\n### Key Architectural Comparison\n| Coordination Engine | Node Primitive | Watch Model | Consensus Protocol | Production Tooling |

:---	:---	:---	:---	:---
etcd	Leases with atomic Compare-and-Swap	Revision stream watches	Raft consensus	Kubernetes Control Plane
HashiCorp Consul	Sessions with Key Locks	Long-polling HTTP blocking queries	Raft consensus	Nomad, Vault

How do you design a real-time distributed leader election mechanism using Apache ZooKeeper or etcd?

Continue Learning on Being SDE