AWS RDS Interview Questions
Core Concepts
Q: What is RDS? Multi-AZ vs Read Replica?
RDS (Relational Database Service) is managed relational databases — AWS handles patching, backups, failover.
Supported engines: MySQL, PostgreSQL, MariaDB, Oracle, SQL Server, Aurora (MySQL/PostgreSQL compatible).
Multi-AZ (High Availability):
•Synchronous replication to standby in another AZ
•Automatic failover in 1-2 minutes (DNS changes, no manual intervention)
•Standby is NOT readable — only for failover
•Use for: production databases, HA requirement
Read Replica (Scalability):
•Asynchronous replication
•Readable (offload read queries from primary)
•Can be promoted to standalone (for DR)
•Up to 5 replicas per primary (MySQL/MariaDB/PostgreSQL)
•Can be cross-region (for geo-distributed reads and DR)
Q: What is Aurora? How is it different from standard RDS?
Aurora is AWS's proprietary cloud-native database — MySQL and PostgreSQL compatible but with significant architectural improvements.
Key Aurora differences:
•6-way replication across 3 AZs (storage layer) — survives 2 AZ failures
•Up to 15 read replicas (vs 5 for standard RDS)
•Failover in ~30 seconds (vs 1-2 mins for Multi-AZ RDS)
•Storage auto-scales in 10GB increments up to 128TB
•Aurora Global Database: single primary, up to 5 secondary regions (sub-second replication)
•Aurora Serverless: auto-scales capacity (good for unpredictable workloads)
When Aurora over standard RDS: Production workloads needing high availability, high throughput, or global scale.
Q: RDS backup and recovery.
Automated backups: Daily snapshots + transaction logs. Retention: 1-35 days. Point-in-time recovery to any second within retention period.
Manual snapshots: User-initiated. Stored until deleted. Persist after RDS deletion.
Cross-region backups: Copy automated backups to another region (DR).
RTO/RPO:
•Multi-AZ: RTO ~1-2 min, RPO near-zero (synchronous)
•Aurora: RTO ~30s, RPO near-zero
•Single-AZ + backups: RTO hours (restore), RPO = last backup
Q: RDS security best practices.
1.Private subnet: Never in public subnet. Use SG to allow only app servers.
2.Encryption at rest: Enable KMS encryption (required for compliance).
3.Encryption in transit: SSL/TLS for connections. Enforce with parameter group (require_secure_transport = ON).
4.IAM Database Authentication: Use IAM roles instead of passwords for MySQL/PostgreSQL connections.
5.Secrets Manager: Rotate database credentials automatically.
6.VPC Security Groups: Only allow inbound from app security group, specific port.
7.No public accessibility: Disable public access.
Q: RDS Performance Insights and optimisation.
Performance Insights: Visualise DB load by wait events, SQL queries. Identify top waits (CPU, I/O, locks).
Optimisation:
•Identify slow queries: Performance Insights top SQL, or pg_stat_statements, slow_query_log
•Add appropriate indexes
•Connection pooling: RDS Proxy (serverless/Lambda use cases)
•Read replicas for read-heavy workloads
•Right-size instance class (use Performance Insights before sizing up)
•Parameter groups: tune max_connections, shared_buffers, etc.
RDS Proxy: Pooled connection management. Reduces overhead for Lambda functions (which create new connections per invocation).
Revision Notes
RDS: managed relational DB. MySQL/PostgreSQL/Aurora/Oracle/SQL Server.
MULTI-AZ: Sync replication, auto-failover (1-2min). Standby NOT readable.
READ REPLICA: Async replication, readable. Up to 5. Promote for DR.
AURORA vs RDS:
6-way replication (3 AZs) | 15 read replicas | 30s failover
Auto-scaling storage | Aurora Global DB (sub-second cross-region)
BACKUP: Automated (1-35 day retention) + Manual snapshots
Point-in-time recovery within retention window
SECURITY:
Private subnet | Encryption at rest (KMS) | SSL in transit
IAM auth | Secrets Manager rotation | No public access
RDS PROXY: connection pooling (Lambda use case)
Performance Insights: identify top wait events and slow queries
