Skip to content
Certification Guide

SAA-C03 · AWS Certified Solutions Architect – Associate

SAA-C03 Quick Reference

Exam guide v1.1

Domain 1: Design Secure Architectures

1.1 Secure access

Decision rule: Give people and workloads temporary credentials through roles, grant the least they need where the access happens, and fence every member account with SCPs, remembering that an explicit deny anywhere always wins.

Evaluation ruleResult
Any matching explicit DenyDenied, whatever allows exist
Same accountIdentity policy or resource policy must allow
Cross-accountCaller's identity policy and owner's resource or trust policy allow
SCP, RCP, permissions boundaryOnly cap permissions; never grant
Limit or behaviorValue
Role maximum session duration1 to 12 hours, default 1 hour
Role chaining session1 hour maximum
SCPs attached per root, OU or account10
SCP document size10,240 characters
RCPs attached per root, OU or account5, including the automatic RCPFullAWSAccess policy
IAM groups per userGroups can't be nested; they can contain only users
Access keys per IAM user2
Root user MFA devicesUp to 8
SCPs and RCPs in the management accountNo effect
SCPs and service-linked rolesNo effect
If the question says…Reach for…
Workforce sign-in to many accountsIAM Identity Center with the company IdP or AD
No one in member accounts may…SCP on the OU or root
New multi-account setup with guardrailsAWS Control Tower
Workload calls AWS APIsIAM role (instance profile or execution role)
Third party needs accessCross-account role with an external ID
Share a resource with the whole organizationResource policy with aws:PrincipalOrgID
Delegate role creation safelyPermissions boundary

1.2 Secure workloads and applications

Decision rule: Put each tier in the most private subnet that still works, let stateful security groups that reference each other carry the allow rules, keep network ACLs for subnet-wide denies, reach AWS services through endpoints instead of the internet, and stop exploits with AWS WAF and floods with AWS Shield.

Filter or controlBehavior
Security groupStateful, allow only, instance level; source can be another security group
Network ACLStateless, allow and deny, subnet level; rules checked in ascending number, first match wins
Network ACL return trafficNeeds rules for the client's ephemeral ports; the range depends on the client OS (32768-61000 on many Linux kernels, 49152-65535 on Windows Server 2008 and later)
Default network ACLAllows all traffic; a custom network ACL denies everything until rules are added
Security group rule quota60 inbound and 60 outbound per group (default); 5 groups per network interface
Network ACL rule quota20 inbound and 20 outbound per ACL (default); up to 40 each on request
NAT gatewayOutbound only for private subnets; 5 Gbps, scales to 100 Gbps; 55,000 connections per destination per IP
Gateway endpointsS3 and DynamoDB only; route-table entry; no additional charge
Interface endpointsMost AWS services through PrivateLink; one network interface per chosen subnet; security group applies
Parameter StoreStandard: 4 KB, 10,000 per account per Region; advanced: 8 KB, 100,000 per account per Region
Secrets ManagerUp to 65,536 bytes per secret; RDS managed master password rotates every 7 days by default
Site-to-Site VPNTwo IPsec tunnels per connection; 1.25 Gbps per standard tunnel, 5 Gbps large bandwidth tunnel on TGW or Cloud WAN
Direct Connect encryptionNot encrypted by default; MACsec on 10, 100 and 400 Gbps dedicated connections at selected locations
AWS WAF rate-based ruleEvaluation window 60, 120, 300 or 600 seconds (default 300); limit minimum 10
Shield StandardAutomatic at no additional charge; layer 3 and 4 DDoS protection
If the question says…Reach for…
Only the app tier may reach the databaseDatabase security group that names the app tier's group
Block one IP range from a subnetNetwork ACL deny with a lower number than the allows
Private instances download updatesNAT gateway in a public subnet, private route to it
Private access to S3 or DynamoDBGateway endpoint plus a bucket policy with aws:SourceVpce
Private access to other AWS servicesInterface endpoint in one subnet per Availability Zone
SQL injection or bad bots in HTTP requestsAWS WAF web ACL with managed rule groups on CloudFront, ALB or API Gateway
Per-IP request floodAWS WAF rate-based rule
Large DDoS with expert help and cost protectionAWS Shield Advanced
Database password rotated automaticallySecrets Manager, managed rotation for RDS
Configuration values, AMI IDsParameter Store standard tier
App users sign in, then reach AWS resourcesCognito user pool, then identity pool
Admin shell without SSH ports or bastionsSystems Manager Session Manager
Encrypted connection to a data center, quick setupSite-to-Site VPN
Private circuit that must be encryptedDirect Connect with MACsec, or a VPN over Direct Connect
Detect compromised credentials or cryptominingAmazon GuardDuty
Find personal data in S3Amazon Macie
Find CVEs on EC2, ECR or LambdaAmazon Inspector

1.3 Data security controls

Decision rule: Match each control to the failure it stops: a customer managed KMS key with a key policy you own for encryption, ACM certificates for TLS, Block Public Access before any bucket policy, Object Lock for deletes that must never happen, and AWS Backup with vault lock plus a copy in another Region or account for recovery.

Control or settingBehavior
AWS managed key (aws/s3)AWS owns the key policy and yearly rotation; you cannot change its properties or schedule its deletion; you cannot use it directly in cryptographic operations
Customer managed keyYour key policy; optional automatic rotation (default every year, custom 90 to 2,560 days); disable; deletion waits 7 to 30 days
KMS key policyMust allow access; IAM policies alone cannot grant KMS access the key policy does not allow
KMS Encrypt payloadUp to 4,096 bytes; larger data uses envelope encryption with GenerateDataKey
Default bucket encryptionSSE-S3 is the base level for every bucket; SSE-KMS with no named key uses aws/s3
RDS encryptionSet only at creation; encrypt an unencrypted instance by copying its snapshot with encryption turned on, then restoring from the encrypted copy
ACM public certificateNo charge for non-exportable public certificates on integrated services; exportable certificates are charged at issuance and renewal (see ACM pricing); renews itself while DNS validation stays in place
ACM for CloudFrontCertificate must be requested in US East (N. Virginia)
ACM email validationBeing deprecated for new public certificates and renewals throughout 2027; move to DNS validation (ACM docs)
S3 Block Public AccessFour settings, at account or bucket level; refuses public policies and ACLs
S3 Object Lock governance modeBlocks normal deletes; users with s3:BypassGovernanceRetention can remove it
S3 Object Lock compliance modeNo user, including root, can shorten or remove retention before it ends
S3 legal holdNo end date; independent of retention; stays until set to off
S3 Object Lock prerequisitesVersioning is required; replicating locked objects needs Object Lock on the destination
S3 ReplicationNeeds versioning on both buckets; copies overwrites and, if enabled, delete markers; version-specific deletes are not replicated, so it is not a backup on its own
AWS Backup vault lock complianceImmutable after the grace time (cooling-off period you set, at least 72 hours); sets minimum and maximum retention for jobs into the vault
Multi-Region KMS keySame key ID and key material in several Regions (KMS docs)
If the question says…Reach for…
Disable the key or control its key policyCustomer managed KMS key
Another account must decryptKey policy allows the other account's role; its IAM policy allows the KMS actions
No public bucket, even with future bucketsAccount-level S3 Block Public Access
Find personal data or public bucketsAmazon Macie
Nobody, not even root, may delete for seven yearsS3 Object Lock, compliance mode, seven-year retention
Open-ended preservation orderS3 Object Lock legal hold
Free HTTPS certificate that renewsACM public certificate, DNS validation, on the ALB or CloudFront
Backups that no admin can deleteAWS Backup vault lock, compliance mode, plus copies to another account
Key scheduled for deletion by mistakeKMS CancelKeyDeletion before the waiting period ends (the key is left Disabled), then EnableKey (KMS docs)

Domain 2: Design Resilient Architectures

2.1 Scalable and loosely coupled architectures

Decision rules:

  • Messaging (2.1a): Name the shape before the service: work done once is a queue (SQS), a copy for every subscriber is a topic (SNS), routing by content or AWS/SaaS events is an event bus (EventBridge), ordered steps with state are a workflow (Step Functions).
  • Compute (2.1b): Pick the most managed compute whose limits the workload fits: Lambda for short event-driven code, Fargate for containers without servers, EC2-backed ECS or EKS for GPUs or host control; EKS only when the stem says Kubernetes.
  • Scaling (2.1c): Scale out stateless copies behind the right load balancer, then cache whatever repeats: CloudFront at the edge, the API Gateway cache for repeated responses, ElastiCache for hot data and sessions, read replicas for read-heavy databases.
Limit or behaviorValue
SQS message sizeUp to 1 MiB
SQS retention1 minute to 14 days, default 4 days
SQS visibility timeout0 seconds to 12 hours, default 30 seconds; for Lambda, at least 6x the function timeout
SQS FIFO throughput (no high-throughput mode)300 transactions per second per partition per API action without high-throughput mode; with batches of 10, up to 3,000 messages per second per API action
SQS FIFO deduplication interval5 minutes
SQS long polling waitUp to 20 seconds
EventBridge targets per rule5
Step Functions durationStandard up to 1 year (exactly-once); Express up to 5 minutes (async at-least-once, sync at-most-once)
API Gateway account throttle10,000 requests per second with a 5,000 burst, per account per Region, shared across HTTP, REST and WebSocket APIs
API Gateway REST integration timeout50 ms minimum; 29 seconds by default, which can be raised above 29 seconds (may lower the Region throttle quota)
API Gateway cache TTLDefault 300 seconds, maximum 3,600 seconds (REST APIs only)
Lambda timeout and memory15 minutes; 128 MB to 10,240 MB
Lambda concurrency1,000 per account per Region by default (adjustable)
Lambda payload6 MB synchronous, 1 MB asynchronous
Lambda package250 MB unzipped (.zip); 10 GB container image
CloudFront default TTL86,400 seconds (1 day) when the origin sends no caching headers
If the question says…Reach for…
Buffer spikes, retry laterSQS standard queue + DLQ
Strict order, no duplicatesSQS FIFO with a message group ID
Same event to several systemsSNS fan-out to one SQS queue each
Route AWS service or SaaS events by contentEventBridge rule
Multi-step process with waits and retriesStep Functions Standard
Existing JMS, AMQP or MQTT appAmazon MQ
Partners send files over SFTPAWS Transfer Family into S3
Over 15 minutes, containerized, no serversECS on Fargate
Existing Kubernetes or Helm toolingAmazon EKS
GPUsECS or EKS on EC2 capacity
No cold starts on a Lambda APIProvisioned concurrency
HTTP path or host routingApplication Load Balancer
TCP or UDP, static IPsNetwork Load Balancer
Global users, shared contentCloudFront
Sessions lost on scale-inSessions in ElastiCache or DynamoDB
Read-heavy databaseRead replicas (details in 3.3)

2.2 High availability and disaster recovery

Decision rules:

  • Strategy (2.2): Match the DR strategy to the stated RPO and RTO and choose the lowest one that meets both. Multi-AZ covers a data center failure; a second Region is needed only for a Region-level requirement.
  • Failover: Use health-checked failover routing in Route 53 and Multi-AZ failover for databases. Keep failover on data plane operations; scaling and weighted changes are control plane actions.
  • Standby readiness: The recovery Region must be able to scale to production capacity, so raise its service quotas and keep its infrastructure defined as code.
DR strategyRPORTOWhat runs in the recovery Region
Backup and restoreHours (minutes with PITR)Up to 24 hoursNothing; backups only
Pilot lightMinutesTens of minutesReplicated data; app servers switched off
Warm standbySecondsMinutesFull stack at reduced capacity
Multi-site active/activeNear zeroPotentially zeroFull stack serving traffic
Limit or behaviorValue
RDS Multi-AZ DB instance failoverTypically 60 to 120 seconds; standby is in another AZ and cannot serve reads
RDS Multi-AZ replicationSynchronous to the standby
Aurora global database failoverReplication lag typically under a second; secondary promoted typically under a minute per the FAQ (the user guide frames RTO in minutes)
JVM DNS cache TTL for RDS endpointsRecommended no more than 60 seconds (RDS docs)
RDS Proxy proxies per account20 per account per Region by default (adjustable through Service Quotas)
RDS Proxy endpointsDefault endpoint spans two AZs; up to 20 additional endpoints per proxy
RDS Proxy placementSame VPC as the database; the proxy itself is not publicly accessible
If the question says…Reach for…
Survive an AZ failure, no Region requirementMulti-AZ (Auto Scaling across AZs, Multi-AZ database)
Lowest cost, RPO in hoursBackup and restore with cross-Region backup copies
RPO in minutes, core database always replicatedPilot light
Standby must serve reduced traffic immediatelyWarm standby
Every Region serves users, RTO near zeroMulti-site active/active
Cross-Region database failover, typically under a minute per the FAQ (user guide frames RTO in minutes)Aurora global database
DNS failover between Regions with health checksRoute 53 failover routing
Same database endpoint after failover, no app changeRDS Multi-AZ
Connection spikes or "too many connections" during failoverRDS Proxy in front of the database
Recovery Region drifted after hand patchesImmutable infrastructure: replace from a tested image
Data must survive loss of one Availability ZoneS3 or a Regional EFS file system (stored across AZs)
Restore an EBS volume in another zone or RegionEBS snapshot, copied to the recovery Region

Domain 3: Design High-Performing Architectures

3.1 High-performing and scalable storage

Decision rule: choose the storage type from how the data is reached (object over an API, a shared file mount, or a disk on one instance), then tune that service to the number in the stem.

Limit or behaviorValue
S3 request rate per partitioned prefixPer prefix: 3,500 PUT/COPY/POST/DELETE and 5,500 GET/HEAD per second; above that, spread requests across prefixes
S3 single PUTUp to 5 GB; multipart upload recommended from 100 MB
S3 multipart uploadUp to 10,000 parts of 5 MiB to 5 GiB
EBS gp33,000 IOPS and 125 MiB/s included; up to 80,000 IOPS and 2,000 MiB/s; 1 GiB to 64 TiB
EBS io2 Block ExpressUp to 256,000 IOPS and 4,000 MiB/s; average latency under 500 microseconds
EBS st1 / sc1Up to 500 / 250 MiB/s; HDD, cannot be boot volumes
EBS Multi-Attachio1 and io2 only; up to 16 Nitro instances in one Availability Zone
EBS Elastic VolumesChange size, type, IOPS, throughput while attached; size can only grow
EFS latency (Standard)About 1 ms reads, 2.7 ms writes (Regional, General Purpose, best case); Windows EC2 cannot mount
EFS modesGeneral Purpose (default, recommended); Elastic (recommended for spiky or unpredictable workloads); Bursting (scales with storage); Max I/O previous gen
Decision guide
If the question says…Reach for…Because…
503 Slow Down from S3More prefixes, retriesRequest rate scales per prefix
Global users upload to one bucketS3 Transfer AccelerationEdge locations shorten the long-distance path
Single-digit ms object access next to computeS3 Express One ZoneLowest-latency S3 class, one AZ
One instance, a database, set IOPSEBS gp3IOPS and throughput set apart from size
Sub-ms latency or beyond gp3 limitsEBS io2 Block ExpressHighest IOPS and lowest latency per volume
Rebuildable cache or scratch, fastest local diskInstance storeHost-attached, lost on stop or terminate
Many Linux servers share files across AZsAmazon EFSRegional NFS file system that scales on its own
Windows SMB with Active DirectoryFSx for Windows File ServerNative Windows file system
HPC or ML on an S3 datasetFSx for Lustre linked to S3Sub-ms latency, very high throughput
NFS, SMB and iSCSI from one systemFSx for NetApp ONTAPMultiprotocol with ONTAP features
On-prem files should land in S3S3 File GatewayNFS or SMB in front, objects behind, local cache
On-prem iSCSI volumes backed by AWSVolume Gateway (cached or stored)Cached: primary in AWS. Stored: primary local
Tape backup softwareTape GatewayVirtual tape library over iSCSI
Migrate a NAS once or on a scheduleAWS DataSyncManaged transfer, not ongoing access

Storage classes and lifecycle rules for cost: see 4.1.

3.2 High-performing and elastic compute

Decision rule: pick the compute shape from how the work arrives and the instance family from the resource that runs out first, then let Auto Scaling follow a metric that grows with load per instance, not a raw total.

Limit or behaviorValue
EC2 type nameSeries letter + generation + options (g Graviton, n network, d instance store) + size
T family (standard mode)Held at baseline CPU once credits are spent; T8i, T4g, T3a, T3 launch as unlimited by default
Default EC2 metricsEvery 5 minutes; detailed monitoring gives 1 minute; memory needs the CloudWatch agent
Simple scaling default cooldown300 seconds; target tracking and step scaling use instance warmup instead
Default instance warmupOptional group setting; AWS suggests starting at 300 seconds if unsure
Predictive scalingNeeds 24 hours of data, uses up to 14 days, forecasts 48 hours; scales out only
Warm pool instance statesStopped, running or hibernated
Default health checkEC2 status checks only; turn on ELB health checks behind a load balancer
Cluster placement groupOne AZ; no burstable instances; 10 Gbps single flow inside versus 5 Gbps outside
Partition placement groupUp to 7 partitions per AZ
Spread placement groupUp to 7 running instances per AZ per group
Lambda memory128 MB to 10,240 MB; CPU scales with memory (1,769 MB = one vCPU); 15-minute maximum
Decision guide
If the question says…Reach for…Because…
CPU-boundC familyMost vCPU per GiB of memory
Memory-boundR or X familyMost memory per vCPU
Training, inference, graphicsP, G, TrnGPUs or ML chips
High local disk IOPSI or D familyLarge instance store
Low average CPU with burstsT familyBaseline plus CPU credits
Load follows a metricTarget trackingAuto Scaling manages the alarms
SQS workersTarget tracking on backlog per instanceFalls as workers are added
Peak at a known timeScheduled actionCapacity before the load
Daily or weekly pattern, slow bootPredictive scaling plus a dynamic policyLaunches ahead; dynamic policy scales in
Long boot, unpredictable spikesWarm poolPre-initialized instances join fast
Tightly coupled HPCCluster placement group, EFALowest latency between nodes
HDFS, HBase, CassandraPartition placement groupRack isolation per partition
A few critical instancesSpread placement groupDistinct hardware per instance
Queued jobs that run to completionAWS BatchManaged job queues and compute environments
CPU-bound Lambda functionMore memoryCPU is allocated in proportion to memory

Lambda, Fargate, ECS and EKS: see 2.1b. Spot, Savings Plans and rightsizing for cost: see 4.2.

3.3 High-performing databases

Decision rule: choose the database by its access pattern, scale relational reads with replicas and a cache, and scale DynamoDB by spreading traffic across partition key values rather than by raising table capacity.

Limit or behaviorValue
RDS read replicas (MySQL, MariaDB)Up to 15 per source in one Region; asynchronous; watch ReplicaLag
Aurora ReplicasUp to 15 per cluster; share the cluster volume; lag typically tens of milliseconds within a Region
Aurora reader endpointBalances connections, not individual queries; writes need the writer endpoint
Aurora storageGrows in 10 GB steps up to 256 TiB on Aurora MySQL 3.10 and later (128 TiB before 3.10)
RDS storage autoscalingUp to a maximum storage threshold; can't decrease storage; limited to 4 storage modifications per 24 hours
gp3 baseline (Db2, MariaDB, MySQL, PostgreSQL)3,000 IOPS and 125 MiB/s below 400 GiB; 12,000 IOPS and 500 MiB/s from 400 GiB. Oracle: from 200 GiB. SQL Server: no threshold
io2 Block ExpressUp to 256,000 IOPS (io2 Block Express storage)
DynamoDB per partition3,000 read units and 1,000 write units per second
DynamoDB item size400 KB maximum
DynamoDB GSIEventually consistent only; sparse; up to 20 per table
DynamoDB LSIDefined at table creation; same partition key; strongly consistent reads; up to 5 per table
DynamoDB on-demandAbsorbs up to double the previous peak instantly; new tables start with 12,000 read and 4,000 write units per second; warm throughput is configurable
DAXItem cache TTL default 5 minutes; up to 10 read replicas per cluster; strongly consistent reads not cached
ElastiCache replicasValkey or Redis OSS: up to 5 replicas per shard; Memcached: no replication
Decision guide
If the question says…Reach for…Because…
Reads slow the relational writerRDS read replicas or Aurora ReplicasReads move off the writer
Many read connections, no replica namesAurora reader endpointFollows replicas as they change
Repeated identical query resultsElastiCache with lazy loading and TTLAnswers repeats; strategy in 2.1c
Sorted sets, lists, or replicated cacheValkey or Redis OSSData structures and replicas
Microsecond DynamoDB reads, eventual is fineDAXCaches items and queries
One hot partition key value throttledHigher-cardinality partition key or write shardingPer-partition limit applies to each key
Query by a non-key attributeGlobal secondary indexOwn key; eventually consistent
Unannounced bursts, no forecastDynamoDB on-demandScales with traffic
Storage fills during a large loadAllocate storage before the loadAutoscaling can't keep pace
gp3 IOPS baseline too lowGrow gp3 to 400 GiB or more (200 GiB on Oracle), or io2 Block ExpressBaseline rises to 12,000 IOPS (not RDS for SQL Server)
Database must move to a different engineSCT for schema and code, DMS for dataHeterogeneous: schema converted first
Same engine, new homeHomogeneous migration (native tools or DMS)Schema and SQL carry over

Failover, Multi-AZ and RDS Proxy: see 2.2. Backups and encryption: see 1.3. Database cost and Aurora Serverless v2 pricing: see 4.3.

3.4 High-performing and scalable network architectures

Decision rule: pick the network path by where the traffic starts and where it must go: Global Accelerator for global users who need fixed IPs, a transit gateway once VPCs outnumber a handful, Direct Connect for steady high bandwidth, and an endpoint to keep AWS service traffic on a short private path.

Limit or behaviorValue
Global Accelerator static IPsTwo IPv4 addresses per accelerator (four with dual-stack); endpoints are ALB, NLB, EC2 or Elastic IP
Transit gateway attachmentsUp to 5,000 per transit gateway (adjustable); one attachment per VPC
Transit gateway bandwidthUp to 100 Gbps per VPC attachment per AZ, each direction
Transit gateway MTU8500 bytes between VPCs, Direct Connect and peering attachments; VPN attachments accept packets up to 1500 bytes, but the usable Site-to-Site VPN tunnel MTU is 1446 bytes (MSS 1406); no jumbo frames
VPC peeringOne-to-one, not transitive; no overlapping CIDR blocks; 9001 MTU in a Region, 8500 across Regions
Site-to-Site VPN tunnelStandard up to 1.25 Gbps; Large Bandwidth Tunnel up to 5 Gbps on a transit gateway or Cloud WAN only, not in five Regions (ap-southeast-4, ca-west-1, eu-central-2, il-central-1, me-central-1)
VPN aggregateECMP needs dynamic routing; two Large Bandwidth connections with four tunnels reach 20 Gbps
Accelerated VPNTransit gateway attachments only; chosen at creation, not switched on later
Direct Connect dedicated ports1, 10, 100 or 400 Gbps
Direct Connect hosted ports50 Mbps to 25 Gbps (partner-provisioned); traffic policing drops bursts above the rate
Gateway endpointsS3 and DynamoDB only; route table entry to the prefix list
Interface endpoint throughput10 Gbps per AZ by default, scaling to 100 Gbps per AZ; MTU 8500; endpoints don't send the ICMP messages Path MTU Discovery needs
Single-flow bandwidth (EC2)5 Gbps outside a cluster placement group; 10 Gbps inside; ENA Express 25 Gbps single flow on supported instances
Internet multi-flow (EC2)Through an internet or local gateway: fewer than 32 vCPUs capped at 5 Gbps; more than 32 vCPUs capped at 50 percent of the instance bandwidth
ENA and EFAENA up to 100 Gbps on supported types (for example C5n); EFA adds OS-bypass for HPC and ML
Decision guide
If the question says…Reach for…Because…
Global users, fixed IPs, TCP or UDP, fast regional failoverGlobal AcceleratorTwo static IPs, edge entry and health-based endpoint choice
Cacheable HTTP content, global usersCloudFront (see 2.1c)Caching at the edge; Global Accelerator never caches
Two or three VPCs, direct link onlyVPC peeringDirect and cheap, but not transitive
Many VPCs, or VPCs plus on-premises through one hubTransit gatewayTransitive routing with route tables per segment
Steady high bandwidth, consistent latencyDirect Connect (dedicated, or a LAG)Private circuit at 1 to 400 Gbps per port
Fast setup over the internet, modest bandwidthSite-to-Site VPNTwo tunnels per connection, 1.25 Gbps each
VPN above 1.25 Gbps on a transit gatewayLarge Bandwidth Tunnels, then ECMP5 Gbps per tunnel; dynamic routing adds connections
Private S3 or DynamoDB from the VPCGateway endpointRoute table entry, no interface to size
Other services, or on-premises or peered callersInterface endpointPrivate IP in your subnets, reachable over VPN and Direct Connect
One flow above 5 Gbps in one RegionENA Express25 Gbps single flow on supported instances
Tightly coupled HPC or ML nodesEFA in a cluster placement groupOS-bypass with Libfabric for MPI and NCCL traffic

Route 53 routing is in 2.2; placement groups are in 3.2; data transfer and NAT cost are in 4.4.

3.5 High-performing data ingestion and transformation

Decision rule: match the ingestion path to the arrival pattern and the consumer: Kinesis Data Streams for ordered real-time reads, Firehose for buffered delivery to S3, MSK when the team already speaks Kafka, and Glue, EMR or Athena for transforms, with Parquet and partitions deciding how fast the data is read later.

Limit or behaviorValue
Kinesis shard, provisioned writes1 MB per second or 1,000 records per second per shard
Kinesis shard, reads2 MB per second and 5 GetRecords calls per second per shard, shared by standard consumers
Enhanced fan-out2 MB per second per registered consumer per shard; up to 20 consumers per stream (50 on On-demand Advantage, in listed Regions only)
Kinesis on-demand scalingUp to double the peak write rate of the previous 30 days; a larger spike can still throttle
Kinesis capacity mode switchTwice per 24 hours, between on-demand and provisioned
Kinesis PutRecordsUp to 500 records per call; HTTP 200 can still carry failed entries, each with an ErrorCode to retry
Firehose buffer to S3Size 1 to 128 MB (default 5); interval 0 to 900 seconds (default 300); first condition met triggers
Firehose record format conversionJSON to Parquet or ORC; S3 is the only destination; buffer size at least 64 MB; Glue table is the schema
Glue worker G.1X1 DPU: 4 vCPUs and 16 GB memory; Auto Scaling needs Glue 3.0 or later and does not support Standard DPUs
Athena partition projectionComputes partitions from table properties, avoiding a Glue Data Catalog partition lookup per query
DataSync multiple agentsUp to four Basic and four Enhanced mode agents per location; no high availability, all agents must be online
Decision guide
If the question says…Reach for…Because…
Several real-time consumers, one needing dedicated read throughputKinesis Data Streams with enhanced fan-out2 MB per second per shard for each registered consumer
Unpredictable traffic, no shard planningKinesis Data Streams on-demandScales to double the 30-day peak; still size for known spikes
Files in S3 with no consumer codeFirehose to S3, buffer tuned for file sizeBatching and retries handled; larger buffer, larger files
JSON stream that analysts query with SQLFirehose Parquet conversion, Glue schemaColumnar, compressed files land in S3
Existing Apache Kafka code, topics or clientsAmazon MSKKafka APIs kept; no client rewrite
Managed Spark ETL that reprocesses only new filesGlue ETL with job bookmarksManaged workers, Auto Scaling, skip processed data
Custom big data frameworks or cluster controlAmazon EMR on EC2 or EKSYou choose the cluster shape and frameworks
SQL over S3 data with no ETLAthena with Parquet and partition projectionColumn pruning and partition pruning cut bytes scanned
One place to restrict tables, columns or S3 pathsLake Formation with a registered locationTable and column permissions across Athena, Glue and EMR
Applications write files over NFS or SMB, and the files must reach S3AWS Storage Gateway (S3 File Gateway)Keeps the file share on premises and copies each file into S3
Dashboards and charts over the S3 lake, with no codeAmazon Quick, with Athena as the sourceVisualizes queried data; it does not ingest or transform it
Producers in a VPC write to Kinesis privatelyInterface VPC endpoint, with an IAM role per producerPrivateLink keeps traffic on the Amazon network; the role limits the streams

Domain 4: Design Cost-Optimized Architectures

4.1 Cost-optimized storage

Decision rule: pay for the access pattern you actually have: match each dataset's read frequency and tolerable retrieval time to the cheapest tier that still meets them, then let lifecycle rules move and expire data for you.

S3 storage classReads aboutFirst byteMin durationMin billable size
S3 StandardMore than monthlyMillisecondsNoneNone
S3 Intelligent-TieringUnknown or changingMilliseconds (archive tiers: restore)NoneUnder 128 KB not tiered
S3 Standard-IAMonthlyMilliseconds30 days128 KB
S3 One Zone-IA (one AZ)Monthly, re-creatableMilliseconds30 days128 KB
S3 Glacier Instant RetrievalQuarterlyMilliseconds90 days128 KB
S3 Glacier Flexible RetrievalYearlyRestore: 1–5 min, 3–5 h, 5–12 h90 daysAbout 40 KB added
S3 Glacier Deep ArchiveLess than yearlyRestore: within 12 h or 48 h180 daysAbout 40 KB added
Limit or behaviorValue
Lifecycle default for small objectsObjects under 128 KB are not transitioned
Lifecycle directionDown the waterfall only; leaving Flexible or Deep Archive needs restore + copy
Intelligent-Tiering movesInfrequent Access after 30 days, Archive Instant Access after 90 days without access
Clean-up actionsNoncurrentVersionExpiration, AbortIncompleteMultipartUpload
Requester PaysRequester pays requests and downloads; owner pays storage
gp3 baseline (in the storage price)3,000 IOPS and 125 MiB/s at any size; gp2 to gp3 online with Elastic Volumes
EBS volume sizeCan grow, can't shrink
st1 / sc1Cheaper HDD for sequential or cold data; never a boot volume
EBS Snapshots Archive90-day minimum; restore up to 72 hours
EFS Archive storage class90-day minimum; Regional file systems with Elastic throughput
EFS One ZoneOne AZ, reduced cost compared to Standard; only for data you can re-create
FSx for Windows File Server leversHDD storage for general shares; deduplication typically saves 50–60% on general-purpose file shares
FSx for NetApp ONTAP leversAuto tiering policy moves cold data to the capacity pool; size SSD for active data
FSx for Lustre leversScratch for temporary processing; not replicated, keep the durable copy in S3
AWS Backup cold storageWarm for at least one week; at least 90 days in cold; not every resource type supports it
DataSync vs Transfer FamilyDataSync copies NFS/SMB/object data over your link; Transfer Family is SFTP/FTPS/FTP/AS2 for partners
Cost allocation tagsMust be activated before they appear in Cost Explorer
Cost Explorer / Budgets / CURExplorer to explore, Budgets to alert (actual or forecast), CUR to analyze (S3 + Athena)

4.2 Cost-optimized compute

Decision rule: rightsize first, then buy for the shape of the usage: Spot for anything that can be interrupted, a Savings Plan or Reserved Instances for the steady baseline, On-Demand for the rest, and nothing at all for capacity that is idle.

Purchase optionMax discountCommitmentFlexibilityReserves capacity
On-DemandNoneNoneAnythingNo
Compute Savings PlansUp to 66%USD per hour, 1 or 3 yearsAny family, size, Region, OS, tenancy; Fargate, LambdaNo
EC2 Instance Savings PlansUp to 72%USD per hour, 1 or 3 yearsSize, OS, tenancy within one family in one RegionNo
Standard RIUp to 72%One configuration, 1 or 3 yearsCan be sold on the RI Marketplace; not exchangeableOnly if zonal
Convertible RIUp to 66%One configuration, 1 or 3 yearsExchangeable; can't be soldOnly if zonal
Spot InstancesUp to 90%NoneInterrupted with a two-minute noticeNo
Capacity ReservationNone itselfNone for immediate start; future-dated has a commitment duration and can carry a cancellation chargeOne instance type in one AZYes, billed if unused
Limit or behaviorValue
Savings Plans and SpotSavings Plans don't discount Spot, and Spot spend doesn't count toward the commitment
Savings Plans and capacityNever reserve capacity; they do discount On-Demand Capacity Reservations
Spot reliabilityMany instance types and AZs, price-capacity-optimized, Capacity Rebalancing
Spot is wrong forStateful, fault-intolerant or tightly coupled work; workloads that need guaranteed capacity
Compute OptimizerOpt-in; 14-day lookback (93 with paid enhanced metrics); 30 hours of EC2 data minimum
Compute Optimizer memoryOnly with the CloudWatch agent (or a supported external tool); it never resizes on its own
HibernationEnabled at launch only; encrypted EBS root sized for RAM; Linux RAM under 150 GiB
Non-production off-hoursScheduled actions set desired capacity (and min) to 0, then back in the morning
Lambda / Fargate / EC2Lambda for spiky or mostly idle; Fargate bills for the task CPU and memory you specify; EC2 for steady use
Discount sharing (Organizations)Commitments apply to the buying account first, then other accounts; management account controls it
Savings Plans budgetAWS Budgets alerts on Savings Plans utilization or coverage below a target
Savings Plan returnWithin 7 days of purchase, same calendar month, hourly commitment $100 or less; yearly return quota applies

4.3 Cost-optimized databases

Decision rule: price the database by its load shape and its data's lifetime: commit to the steady baseline, let on-demand or Serverless v2 carry the unpredictable part, expire data you no longer need, and cache repeated reads so the database can stay smaller.

Billing or storage choiceUse whenKey limit or behavior
DynamoDB on-demandTraffic is unpredictable or newPays per request; costs more per request than well-used provisioned capacity
DynamoDB provisioned with auto scalingTraffic is steady and forecastablePays for provisioned units; reserved capacity lowers the rate on the floor
DynamoDB Standard-IAStorage exceeds 50% of the read and write cost and reads are rareStorage about 60% lower; reads and writes about 25% higher; no reserved capacity
DynamoDB TTLItems expire at a known timeEpoch-seconds number attribute; deletion is background and can lag; filter expired items from reads
Aurora Standard vs I/O-OptimizedCompare both bills; I/O-heavy favors I/O-OptimizedI/O-Optimized has no per-request I/O charge but higher instance and storage rates; pays off when I/O spend exceeds 25% of the bill; switch once every 30 days
Aurora Serverless v2Idle or spiky loadMinimum 0 ACUs enables auto-pause; storage and backups still billed
RDS storage autoscalingGrowth is hard to forecastRaises allocated storage; can't decrease it; allocate before very large loads
Free backup storageKeep retention to what the requirement namesRDS: 100% of provisioned storage per Region per account; Aurora: latest cluster volume size, for automated backups
Multi-AZ standbyAvailability requirement (see task 2.2)Standby instance and storage are billed; it is not a backup
Database Savings PlansCommitted spend across Aurora, RDS, DynamoDB, ElastiCacheUp to 35% off; one-year term; flexible across those services
DynamoDB reserved capacitySteady provisioned baseline, one or three yearsBlocks of 100 capacity units; up to 54% off (1 year) or 77% (3 years) in select Regions; provisioned Standard table class only
RDS reserved DB instancesFixed database configuration for a fixed termOne- or three-year term; one instance type, engine and Region; does not reserve capacity
ElastiCache reserved nodesCache cluster runs at a steady sizeDiscounted rate on pre-purchased nodes; Valkey priced below Redis OSS
Redshift ServerlessBursty warehouse queriesBilled per RPU-hour while queries run; set usage limits; steady warehouse fits provisioned
ElastiCache in front of the databaseThe same reads repeatOnly repeated reads are saved; the cache is itself a running cost

4.4 Cost-optimized network

Decision rule: data transfer is priced by direction and path: keep traffic in one Availability Zone where you can, send private AWS traffic through gateway endpoints, use NAT gateways only for low-volume egress, and move high-volume internet traffic onto CloudFront or Direct Connect.

PathHourly chargePer-GB chargeCost note
Into AWS from the internetNoneNoneFree
Out to the internetNoneTiered per GBMost expensive per-GB path; put high volume behind CloudFront
Same AZ, private IPNoneNoneUsually free over private IPs; co-locate chatty tiers
Between AZs in one RegionNoneBoth directionsBilled for inbound and outbound
Between RegionsNoneOutbound from the source Region onlyDestination inbound has no charge
Gateway endpoint (S3, DynamoDB)NoneNoneAlways use before NAT for these services
Interface endpointPer endpoint per AZPer GB processedCheaper than NAT only at volume
NAT gateway, zonalPer gateway per AZPer GB processed through the gatewayOne per AZ with workloads; NAT plus cross-AZ adds cost
NAT gateway, regionalPer AZ-hour it is configured inPer GB processed through the gatewayExpands across AZs with workloads; HA by default; not for private connectivity
NAT instanceEC2 instance costNormal data transfer, no processing feeSelf-managed; not the production answer
VPC peeringNoneData transfer rate for the pathCheapest link for two or a few VPCs
Transit gatewayPer VPC attachment, per hourPer GB processed; none on data from a peering attachmentPays off when the peering mesh becomes hard to run
Site-to-Site VPNPer connectionInternet data transfer rateLow volume and backup
Direct ConnectPort-hour chargeData transfer out at the DX rate; in is $0.00/GBSteady high volume to on-premises
CloudFrontNone; usage-based pricingOut to viewers; AWS origin to edge is waivedBest path for high-volume public content; AWS origin to edge transfer is free
Global AcceleratorPer accelerator, per hourData transfer plus an incremental charge on top of standard ratesBuy it for latency, static IPs or cross-Region failover, not to save money
Route 53 alias to an AWS resourceHosted zone per monthNone for alias queries to a load balancer, CloudFront, Global Accelerator or S3 websiteA CNAME to a Route 53 record is billed as two queries per lookup
Public IPv4 addressPer address, in use or idleNoneShare with a load balancer or NAT; check the CUR line
TrapCorrect answer
One NAT gateway for workloads in three AZsOne zonal NAT gateway per AZ with workloads, or a regional NAT gateway
NAT gateway carrying S3 or DynamoDB trafficGateway endpoint for S3 or DynamoDB (free)
Interface endpoint for S3 to save moneyGateway endpoint for S3; interface endpoints charge per hour and per GB
NAT instance in production to save moneyManaged NAT gateway per AZ
Site-to-Site VPN for terabytes a monthDirect Connect for steady high-volume hybrid traffic
Transit gateway for two VPCs to save moneyVPC peering (no hourly charge)
Ignoring public IPv4 chargesEvery public IPv4 address is billed per hour, in use or idle

Last reviewed