Skip to main content
Welcome back. Now that you’ve deployed an application and understand how it interacts with a database, this article explains Amazon RDS instance types and how to choose between them. We’ll cover the common instance families, typical use cases, operational trade-offs, and monitoring tips so you can match instance choice to application workload and cost constraints.

​
Overview

Amazon RDS provides multiple instance classes optimized for different workloads and resource profiles. The most common families are:
  • Burstable performance instances (T classes) — cost-effective for workloads with mostly low CPU usage but occasional bursts.
  • General purpose instances (M classes) — balanced CPU, memory, and network; good default for steady workloads.
  • Memory-optimized instances (R classes) — for memory-heavy workloads such as in-memory caches or analytics.
Choosing the right family is a balance between performance requirements and cost. Match instance class to your workload pattern (steady vs. spiky vs. memory intensive) and validate with monitoring metrics over time.

​
Instance family details

General purpose instances
  • Provide balanced CPU, memory, and networking.
  • Recommended for moderate and consistent workloads, such as OLTP or small-to-medium web applications.
  • Cost-effective when you need predictable capacity across multiple resource dimensions.
Burstable performance instances (T classes)
  • Designed for workloads that are mostly idle or low-utilization but occasionally require short CPU bursts.
  • Use a CPU credits model: when CPU utilization is below the baseline, credits accumulate; when utilization exceeds the baseline, credits are consumed to permit bursts.
  • Best for workloads with unpredictable spikes where sustained high CPU is not common.
Memory-optimized instances (R classes)
  • Offer high memory capacity and memory throughput.
  • Ideal for large in-memory caches, high-performance databases, and analytics workloads that are memory-bound.
When an application experiences increased traffic and needs more compute, the instance uses accumulated CPU credits to serve that traffic, enabling short-term higher performance without permanently provisioning a larger instance.
A slide titled "RDS Instances Types" comparing two instance classes. The left box describes Balanced Performance as a compromise between CPU, memory, and network for apps not receiving heavy traffic; the right box describes Burstable Performance, which accumulates CPU credits during low utilization to use during spikes.

​
Example scenario

Consider a coupon-generation service that normally sees low traffic but experiences large spikes when a campaign launches or when a partner integration triggers many requests. For this workload:
  • A burstable instance (T-class) can be economical because it handles short spikes by consuming CPU credits and keeps baseline costs low.
  • If the spikes become frequent or prolonged, re-evaluate and consider a general purpose (M) or memory-optimized (R) instance depending on the bottleneck.

​
Quick decision table

Instance familyTypical use caseWhen to choose
Burstable (T)Low baseline CPU with occasional spikesLow-cost option for unpredictable, short spikes; not for sustained high CPU
General purpose (M)Balanced CPU, memory, networkDefault choice for steady, moderate workloads
Memory-optimized (R)Memory-intensive workloadsLarge caches, in-memory DBs, analytics

​
Practical tips for choosing and validating

  • Start with the family that matches your workload pattern. Use M for steady workloads, T for low baseline with spikes, and R for memory-bound workloads.
  • Monitor these CloudWatch metrics to validate instance sizing:
    • CPUUtilization
    • FreeableMemory
    • ReadIOPS / WriteIOPS
    • DiskQueueDepth / BurstBalance (for burstable storage)
  • If your T-class instance frequently runs out of CPU credits and performance degrades, migrate to an M-class or scale vertically.
  • Consider storage type and IOPS separately — instance class handles compute and memory; storage choices (General Purpose SSD, Provisioned IOPS, Magnetic) affect disk throughput and latency.
Burstable T instances are not suitable for sustained high CPU workloads. If CPU credits are exhausted repeatedly, you will see degraded performance. Monitor credits and CPU metrics and plan a resize or family change if credits run low.

​
Example: list available DB instance classes (AWS CLI)

Use this AWS CLI command to list available RDS instance classes for a specific engine and region: 
aws rds describe-orderable-db-instance-options \
  --engine mysql \
  --query 'OrderableDBInstanceOptions[].DBInstanceClass' \
  --output text | tr '\t' '\n' | sort -u
Adjust --engine to your database engine (postgres, aurora, sqlserver, etc.).
Beyond instance class, tune storage type and provisioned IOPS for I/O-sensitive workloads. Always validate choices with production-like load tests and CloudWatch metrics before committing to long-term provisioning.

​
Summary

  • Use M for balanced, steady workloads.
  • Use T for low baseline, occasional spikes — but monitor CPU credits.
  • Use R for memory-intensive workloads with high memory throughput.
  • Monitor CloudWatch metrics to validate sizing and catch issues early.

Watch Video