Right-size your compute
Right-size, Graviton and idle instances: match spend to what you actually run.
Lessons in this path
- 1 Cost AWS
Migrate EC2 to Graviton (ARM)
Up to 40% better price-performance on ARM, surfaced by Cost Optimization Hub as a recommendation. Validate compatibility, then ride the migration loop.
18 min - 2 Cost AWS
Migrate RDS to Graviton (ARM)
db.m6g/r6g instances deliver ~20% better price-performance than x86 with zero application changes; apply the change in a maintenance window.
16 min - 3 Cost AWS
Right-size EC2 instance
Match instance types to actual workload: stop overpaying for unused capacity.
15 min - 4 Cost AWS
Reclaim idle EC2 instances
Find EC2 instances that are running but doing nothing: stop or terminate them before they keep billing.
14 min - 5 Cost AWS
Upgrade EC2 instances to the newest generation
Each new EC2 generation delivers ~10-20% better price/performance at the same hourly rate, so the same workload runs on fewer or smaller instances. Cost Optimization Hub surfaces the swaps. Apply them.
14 min - 6 Cost AWS
Clean up stopped EC2 instances with attached EBS
A stopped instance costs nothing, but its attached EBS volumes keep billing every hour: find them and decide before the bill compounds.
12 min - 7 Compliance AWS
Remove long-stopped EC2 instances
Security Hub EC2.4 — instances stopped for >30 days are usually abandoned. They keep EBS billing and AMIs/snapshots stale.
11 min - 8 Cost AWS
Stop or delete idle RDS databases
Multi-AZ db.m5.large with 0.3 average connections is paying for nothing: find the idle databases and choose stop, snapshot+delete, or Aurora Serverless.
16 min - 9 Cost AWS
Right-size ECS tasks and Fargate services
Fargate bills for the vCPU and memory you request in the task definition, not what your containers actually use. Over-requesting is pure waste, multiplied by every running task.
15 min - 10 Cost AWS
Right-size Lambda function memory
Memory is the only dial on a Lambda, and it controls CPU and network too. Set it wrong and you either overpay on idle RAM or pay more on slow runs.
14 min - 11 Cost AWS
Right-size Lambda provisioned concurrency
Provisioned concurrency keeps execution environments warm to kill cold starts, but you pay for every warm second whether traffic uses it or not. Set it wrong and you're heating an empty room around the clock.
13 min - 12 Cost AWS
Migrate Lambda functions to Graviton (ARM)
Roughly 20% lower price per GB-second, plus better price-performance that shortens duration: a one-line flip for pure-interpreted runtimes, real build work for native dependencies.
15 min - 13 Cost AWS
Switch DynamoDB tables between on-demand and provisioned capacity
DynamoDB has two billing modes priced an order of magnitude apart, and picking the wrong one per table is one of the quietest, most common ways to overpay for a database that's working perfectly.
14 min - 14 Cost AWS
Convert idle RDS to Aurora Serverless v2
A dev database that sits idle nights and weekends still pays full price 24/7 on provisioned RDS; Aurora Serverless v2 scales capacity down when no one's using it, so you stop renting a box that's mostly asleep.
15 min - 15 Cost AWS
Migrate Aurora to I/O-Optimized
Aurora Standard charges per I/O request; for I/O-heavy clusters that bill is unpredictable and large. I/O-Optimized removes it entirely. Know the breakeven before you switch.
14 min