AWS RDS

RDS

• Relational Database Service, a managed service
• automated provisioning, OS patching
• continuous backups and restore to specific timestamp (Point In Time Restore)
• monitoring dashboards
• read replicas for improved read performance
• Multi AZ setup for DR (Disaster Recovery)
• maintenance windows for upgrades
• scaling capability (vertical and horizontal)
• storage backed by EBS (gp2 or io1)

Storage Auto Scaling

• when RDS detects you are running out of free database storage, it scales automatically
• you have to set Maximum Storage Threshold

Deployment Options

• Multi AZ DB instance
  • creates a primary DB instance with one standby DB instance in a different AZ
  • provides high availability and data redundancy, but the standby replica cannot serve read traffic
  • increased write and commit latency (recommend to use Provisioned IOPS for better performance)
  • auto-failover to the standby replica
• Multi AZ DB cluster
  • creates a DB cluster with a primary DB instance and two readable standby DB instances, with each DB instance in a different AZ
  • provides high availability, data redundancy and increases capacity to serve read workloads
  • semi-synchronous replication, which requires acknowledgment from at least one reader DB instance
  • auto-failover to one of the reader DB instances
• Single DB instance
  • creates a single DB instance with no standby DB instances
• Read replicas
  • main purpose is scalability
  • asynchronous replication, all read replicas are accessible and can be used for read scaling
• Multi-Region
  • main purpose is disaster recovery and local performance
  • asynchronous replication and all regions are accessible and can be used for reads
• Multi-AZ:
  • Convert RDS instance from Single-AZ to Multi-AZ, the following happens:
    • a snapshot of your primary instance is taken
    • a new standby instance is created in a different AZ from the snapshot
    • synchronous replication is configured between primary and standby instances
    • no downtime incurred, but the latency might increase due to the synchronous replication
  • During Multi-AZ failover, the following happens:
    • failover is automatically handled by RDS without administrative intervention
    • RDS simply flips the canonical name record (CNAME) for your DB instance to point at the standby, which is in turn promoted to become the new primary
    • the failover typically completes within one or two minutes
  • The following events would case RDS to initiate a failover to the standby replica:
    • loss of availability in primary AZ
    • loss of network connectivity to primary
    • compute unit failure on primary
    • storage failure on primary

When creating a new RDS database, you can choose the Deployment options from Multi-AZ DB cluster, Multi-AZ DB instance and Multi-AZ DB instance. For existing RDS databases, you can Create new replica and the new replica can be in different regions.

RDS vs RDS Custom

• RDS: entire database and the OS to be managed by AWS
• RDS Custom:
  • managed Oracle and Microsoft SQL Server database with OS and database customisation
  • full admin access to the underlying OS and the database

Security

• data at-rest encryption
  • master and replicas, KMS, enabled at launch time
  • un-encrypted database => DB snapshot => restore as encrypted
  • master is un-encrypted, then replicas cannot be encrypted
• in-flight encryption: TLS-ready by default
• IAM database authentication
• Audit logs
• no SSH except for RDS custom

RDS Proxy

• allows apps to pool and share DB connections established with the database
• improves database efficiency by reducing stress on database resources (CPU & RAM) and minimise open connections (timeout)
• serverless, auto-scaling, highly available (multi AZ)
• never publicly accessible (must be accessed from VPC)
• use case: lambda function (auto scale) => RDS proxy (pool connections) => RDS

Monitoring

• Performance Insights
  • a database performance tuning and monitoring feature that helps you quickly assess the load on your database
  • uses lightweight data collection methods that don't impact the performance of you application, and makes it easy to see which SQL statements are causing the load and why
• Enhanced monitoring
  • you can monitor the operating system of your DB instance in real time
  • you can view all the system metrics and process information for your RDS DB instances on the console
  • main metrics: cpuUtilization, diskIO, loadAverageMinute, memory, network, processList, swap
  • difference between CloudWatch and Enhanced Monitoring
    • CloudWatch gathers metrics about CPU utilisation from the hypervisor for a DB instance
    • Enhanced Monitoring gathers its metrics from an agent on the DB instance
    • the hypervisor only does a small amount of work