Exam Numbers Cheat Sheet
💡 Every limit and default worth memorizing, in one place. Numbers marked ⚠ are ones AWS occasionally revises — sanity-check anything surprising against current docs.
Lambda
| Number | What it is |
|---|
| 15 min | Max execution timeout |
| 10 GB | Max memory (CPU scales with it) |
| 1,000 ⚠ | Default account concurrency (soft limit) |
| 6 MB / 256 KB | Max payload: synchronous / asynchronous |
| 2 retries | Async invocation retries before discard (→ destination/DLQ) |
| 250 MB / 10 GB | Unzipped deployment package / container image max |
| 512 MB – 10 GB | /tmp storage |
SQS / SNS
| Number | What it is |
|---|
| 256 KB | Max message size (both) — larger → Extended Client Library + S3 |
| 4 days / 14 days | Default / max message retention |
| 30 s / 12 h | Default / max visibility timeout |
| 20 s | Max long-polling wait |
| 300 msg/s / 3,000 msg/s ⚠ | FIFO throughput: without / with batching |
| 15 min | Max delivery delay (delay queue) |
| 5 min | Deduplication window (FIFO) |
Kinesis
| Number | What it is |
|---|
| 1 MB/s or 1,000 rec/s | Ingest per shard |
| 2 MB/s | Egress per shard (shared across consumers; per-consumer with Enhanced Fan-Out) |
| 24 h / 365 days | Default / max retention |
| 1 MB | Max record size |
Step Functions
| Number | What it is |
|---|
| 1 year / 5 min | Max duration: Standard / Express |
| 25,000 | Execution history events (→ Distributed Map for more) |
| 256 KB | Max state input/output payload |
DynamoDB
| Number | What it is |
|---|
| 400 KB | Max item size |
| 48 h | TTL deletion can lag this long after expiry |
| 24 h | Streams retention |
S3
| Number | What it is |
|---|
| 5 TB / 5 GB | Max object / max single PUT (multipart beyond) |
| 100 MB | Multipart recommended above this |
| 3,500 / 5,500 | Writes / reads per second per prefix |
| 7 days | Max presigned URL validity (SigV4) |
IAM / STS / KMS
| Number | What it is |
|---|
| 1 h | Default STS session; hard cap for role chaining |
| 12 h | Max role session duration (if configured) |
| 4 KB | Max data for direct KMS Encrypt (→ envelope encryption) |
| 7–30 days | KMS key deletion waiting period |
| 12 h | CodeArtifact auth token max |
| 1 year | Automatic KMS key rotation interval |
CI/CD
| Number | What it is |
|---|
| 60 min / 8 h | CodeBuild default / max build timeout |
| 5 | Max targets per EventBridge rule |
| 24 h | EventBridge retry window before drop/DLQ |
| 1,000 | Beanstalk application version limit (→ lifecycle policy) |
| 10 / 1 | CodeArtifact: max upstreams / max external connections per repo |
ASG / EC2 / ELB
| Number | What it is |
|---|
| 300 s | Default cooldown; default health check grace period |
| 1 h | Default lifecycle hook heartbeat timeout (extend via heartbeat, up to 48 h total) |
| 5 min / 1 min | EC2 metrics: standard / detailed monitoring interval |
| 2 min | Spot interruption notice |
| 1 | HttpPutResponseHopLimit that blocks containers from IMDS |
CloudWatch / Logs
| Number | What it is |
|---|
| 1 s | Highest-resolution custom metric granularity |
| 10 s / 30 s | Fastest alarm evaluation on high-res metrics |
| 100 | Metric filters per log group |
| Never expire | Default log retention (set it explicitly!) |
| 15 months | Metric data retention (rolled up over time) |
Databases / DR
| Number | What it is |
|---|
| < 1 s / < 1 min | Aurora Global Database typical RPO / RTO |
| 15 | Max Aurora read replicas per cluster |
| 35 days | Max automated backup retention (RDS/Aurora) |
| 72 h | Aurora Backtrack max window |
| 5 | RDS read replicas (non-Aurora) ⚠ (15 for some engines now) |
Route 53 / CloudTrail / Misc
| Number | What it is |
|---|
| 30 s / 10 s | Health check: standard / fast interval |
| 3 | Consecutive failures → unhealthy (default) |
| 8 | Max records returned by multivalue answer routing |
| 90 days | CloudTrail Event History retention (fixed) |
| ~7 days | CloudTrail Insights baseline stabilization |
| 15 min | S3 Replication Time Control SLA |