Amazon RDS (Relational Database Service)

RDS provides managed relational databases: MySQL, PostgreSQL, MariaDB, Oracle, and SQL Server. AWS handles provisioning, patching, backups, Multi-AZ, and monitoring. You manage application-level optimization, query performance, and instance sizing.

Engines

Engine	Version	Use Case	License
MySQL	5.7, 8.0	Web apps, SaaS	Open source
PostgreSQL	13, 14, 15, 16	Enterprise, geospatial	Open source
MariaDB	10.2, 10.3, 10.4, 10.5, 10.6, 10.11	MySQL drop-in replacement	Open source
Oracle	19c, 21c	Enterprise (existing Oracle apps)	BYOL or Oracle License
SQL Server	2014, 2016, 2017, 2019, 2022	Windows/.NET apps	BYOL or SQL Server License

Instance Classes

db.t3.micro      → Burstable (2 vCPU, 1 GB)
db.m5.large      → General (2 vCPU, 8 GB)
db.m5.xlarge     → General (4 vCPU, 16 GB)
db.r6g.large     → Memory optimized (2 vCPU, 16 GB)
db.r6g.xlarge    → Memory optimized (4 vCPU, 32 GB)
db.r6g.2xlarge   → Memory optimized (8 vCPU, 64 GB)
db.m5.24xlarge   → High CPU (96 vCPU, 384 GB)

Current gen (use these): T3, M5, M6I, R5, R6I, R6G Previous gen (avoid for new): T2, M4, R4

Creating a DB Instance

Via Console

RDS → Create database → Choose engine → Select use case (Production/Dev) → Configure settings

Via CLI

aws rds create-db-instance \
  --db-instance-identifier my-db \
  --db-instance-class db.t3.micro \
  --engine postgres \
  --engine-version 15.3 \
  --master-username postgres \
  --master-user-password SecretPassword \
  --allocated-storage 20 \
  --storage-type gp3 \
  --vpc-security-group-ids sg-xxxxx \
  --db-subnet-group my-subnet-group \
  --backup-retention-period 7 \
  --multi-az \
  --preferred-backup-window 03:00-04:00 \
  --preferred-maintenance-window sun:04:00-sun:05:00

High Availability (Multi-AZ)

How It Works

Primary (AZ-1) ◄─────── Sync replication ────────► Standby (AZ-2)
     │
     │  Async replication
     ▼
Read Replica (AZ-3)

When Multi-AZ is enabled:

Synchronous replication to standby in different AZ
Automatic failover (< 60 seconds, typically 20-30 seconds)
No application code changes needed (endpoint stays the same)
Standby cannot be used for reads (it’s a hot standby)

Failover Trigger Conditions

AZ outage (primary AZ goes down)
Instance failure (EC2 host hardware/software failure)
Manual failover (aws rds reboot-db-instance --force-failover)
Maintenance events requiring reboot

Read Replicas

Asynchronous replication for read scaling:

# Create read replica
aws rds create-db-instance-read-replica \
  --db-instance-identifier my-db-replica \
  --source-db-instance-identifier my-db \
  --db-instance-class db.t3.micro \
  --no-multi-az

Use Cases

Scale read-heavy workloads
Analytics/reporting queries
Cross-region DR
Major version upgrades (by promoting replica)

Limitations

Asynchronous (some lag, typically < 1 second)
Not automatically promoted on primary failure
Replica lag is visible in CloudWatch (ReplicaLag metric)
Binary log (MySQL) or WAL (PostgreSQL) must be enabled

Automated Backups

# Backup settings (set at creation)
--backup-retention-period 7      # days (0 to disable)
--backup-window 03:00-04:00      # start time
--preferred-backup-window 03:00-04:00

Restoring

# Restore to point in time
aws rds restore-db-instance-to-point-in-time \
  --source-db-instance-identifier my-db \
  --target-db-instance-identifier my-db-restored \
  --restore-time 2024-01-15T10:00:00Z
 
# Restore from snapshot
aws rds restore-db-instance-from-db-snapshot \
  --db-instance-identifier my-db-restored \
  --db-snapshot-identifier my-snapshot

Important: Restoring creates a NEW instance with a new endpoint. You must update your application to point to the new endpoint.

Parameter Groups

Database configuration (like my.cnf or postgresql.conf):

# Create parameter group
aws rds create-db-parameter-group \
  --db-parameter-group-name my-custom-params \
  --db-parameter-group-family postgres15 \
  --description "Custom parameters"
 
# Modify parameters
aws rds modify-db-parameter-group \
  --db-parameter-group-name my-custom-params \
  --parameters '[
    {"ParameterName": "max_connections", "ParameterValue": "200", "ApplyMethod": "pending-reboot"},
    {"ParameterName": "work_mem", "ParameterValue": "4096", "ApplyMethod": "immediate"}
  ]'

Common parameters:

max_connections — max concurrent connections
shared_buffers — PostgreSQL buffer cache
innodb_buffer_pool_size — MySQL buffer pool
log_min_duration_statement — slow query threshold

Option Groups

Add features (Oracle ASM, SQL Server Transparent Data Encryption, etc.):

# Add option to option group
aws rds modify-option-group \
  --option-group-name my-option-group \
  --options '[{"OptionName": "MEMCACHED", "VpcSecurityGroupMemberships": ["sg-xxxxx"]}]'

Enhanced Monitoring

# Enable enhanced monitoring
aws rds modify-db-instance \
  --db-instance-identifier my-db \
  --monitoring-interval 60 \
  --monitoring-role-arn arn:aws:iam::123456789012:role/rds-monitoring-role

Key metrics (CloudWatch vs Enhanced Monitoring):

CloudWatch: CPU, Connections, Disk queue depth, Swap usage
Enhanced: OS-level (processes, memory, I/O per process)

Connecting

# Get endpoint
aws rds describe-db-instances \
  --db-instance-identifier my-db \
  --query 'DBInstances[0].Endpoint'
 
# PostgreSQL
psql -h my-db.xxxxx.us-east-1.rds.amazonaws.com -U postgres -d mydb
 
# MySQL
mysql -h my-db.xxxxx.us-east-1.rds.amazonaws.com -u postgres -p mydb
 
# SSL
psql "host=my-db.xxxxx.rds.amazonaws.com port=5432 dbname=mydb sslmode=require"

Performance Insights

DBA (Database Inspector) for real-time SQL analysis:

# Enable Performance Insights
aws rds modify-db-instance \
  --db-instance-identifier my-db \
  --enable-performance-insights \
  --performance-insights-kms-key-id arn:aws:kms:us-east-1:123456789012:key/xxxxx

View in Console: RDS → Instances → my-db → Performance Insights

Pricing

Component	Cost
db.t3.micro	$0.017/ h r ($ 12/month)
db.m5.large	$0.136/ h r ($ 98/month)
db.r6g.xlarge	$0.252/ h r ($ 181/month)
Multi-AZ	2x instance cost
Read Replica	Same as single instance
Storage (gp3)	$0.08/GB/month
Backup storage	$0.095/GB/month
Data transfer	$0.02-0.09/GB

Storage Types

Type	Description	Use
gp3	General purpose SSD, 3000 IOPS baseline	General workloads
gp2	General purpose SSD, burst to 3000 IOPS	Legacy
io1	Provisioned IOPS (up to 64,000)	High IOPS (64K+)
io2	Provisioned IOPS (up to 256,000)	Highest IOPS
io2 Block Express	Up to 256,000 IOPS, 64 TB	Maximum performance

Limits

Resource	Limit
DB instances per region	40 (soft)
Storage per instance	64 TB
Max databases per instance	MySQL: unlimited, PostgreSQL: unlimited
Max connections	db.t3.micro: 100, db.m5.xlarge: 1000+
Parameter groups per region	50
Read replicas per primary	5 (MySQL/MariaDB), 5 (PostgreSQL)

References

Homepage: https://aws.amazon.com/rds/
Documentation: https://docs.aws.amazon.com/AmazonRDS/
Pricing: https://aws.amazon.com/rds/pricing/

Pricing Examples

Scenario 1: A small production PostgreSQL database (db.m5.large, Multi-AZ). On-Demand: $0.136/ h r \times 2 (M u lt i - A Z) \times 24 \times 30 =$ 195.84/month + storage 100GB × $0.08 =$ 8/month. Total: ~ $204/ m o n t h . Wi t h R D SR eser v e d I n s t an ce (1 ye a r, n o u p f ro n t) :$ 0.083/hr effective = $119.52/ m o n t h + s t or a g e =$ 127.52/month. Savings: 38%.

Scenario 2: A dev PostgreSQL database (db.t3.micro, single-AZ). On-Demand: $0.017/ h r \times 24 \times 30 =$ 12.24/month. Storage 20GB × $0.08 =$ 1.60/month. Total: ~$14/month. Dev instances can use db.t3.micro (covered by free tier for new accounts) or stop/start to avoid charges.

Nuggets & Gotchas

RDS storage auto-scales but only UP — you cannot shrink storage once increased: If you provision 100GB and only use 20GB, you still pay for 100GB. Start with minimal storage and let RDS auto-scale as needed.
RDS parameter groups require reboot to apply many parameters — not all changes are immediate: If you change max_connections or shared_buffers, you typically need to reboot the instance. Check pending-reboot status in parameter group.
MySQL/RDS read replicas use binary log replication — this adds load to the primary: For heavily write-intensive workloads, binary log replication can cause replication lag. Use PostgreSQL’s WAL-based replication if you need less overhead.
RDS automated backups run in the maintenance window — don’t schedule backups during peak hours: The default 30-minute window runs during your configured time. Set it to off-peak (e.g., 3 AM).
You cannot connect to RDS from the internet — it must be in a VPC with proper security groups: If you try to telnet rds-endpoint 5432 from outside AWS and it fails, check your VPC (should be private subnet), security group (allow your IP on port 5432), and subnet routing (should not have IGW, should use NAT for outbound).

cloudnative wiki

Explorer

Amazon RDS