Community Q&A
SAP-C02 — frequently asked questions
Community-vetted answers to 39 common questions about this exam, each traced back to a specific practice question.
Questions from real practice questions
Each Q&A comes from a specific question — follow the link for the full analysisImproving Application Reliability with Auto Scaling and Aurora
Why is adding a second VPC peering connection (B) wrong for reliability?
VPC peering is not a high-availability mechanism. It connects two VPCs but does not provide redundancy against AZ failures or instance crashes within those VPCs.
Can I move an existing EC2 instance to another VPC easily?
No. You cannot directly move a running EC2 instance to a different VPC. You must create a new instance in the target VPC from a snapshot or AMI, which disrupts service.
How Should a Company Design High-Performance Hybrid DNS for cloud.example.com?
Why is a Route 53 outbound resolver wrong for on-premises queries?
An outbound Resolver endpoint forwards DNS queries from VPCs to on-premises resolvers; on-premises systems need an inbound endpoint to query Route 53 private hosted zones.
Does associating the private hosted zone with every VPC reduce performance?
No. Local association lets each VPC answer cloud.example.com directly, avoiding hairpins through the shared services VPC and improving resolution performance.
How to Let a New Account Update AWS Config Under Root SCPs?
Why can’t an allow SCP on a temporary Onboarding OU override the root deny-list SCP?
SCPs are filters, not grants; an explicit deny at the organization root still applies to every child OU and account, so a lower-level allow cannot take effect while the root deny remains.
Why move the root SCP to the Production OU instead of deleting it?
Moving it keeps the same deny-list protections for Production accounts while freeing the new account in Onboarding to update AWS Config; deleting it would remove enforcement for everyone.
How to Share VPC Subnets Across AWS Accounts with RAM?
Why is a transit gateway (A) not needed for sharing subnets across accounts?
A TGW routes traffic between VPCs but does not grant accounts permission to launch resources in another account's subnets; the requirement is delegated subnet use via AWS RAM.
Why must resource sharing be enabled in the Organizations management account?
AWS RAM sharing with an OU requires resource sharing enabled at the organization level; without it, the infrastructure account's resource share cannot target the OU.
Which S3 storage solution archives private documents at the lowest cost?
Why is S3 Glacier Deep Archive not the lowest-cost answer here?
Its per-GB storage is cheaper, but restoring archived objects takes hours and adds retrieval/request costs; without a restore workflow the intranet cannot serve them directly.
Why does S3 One Zone-IA fit data copied from physical media?
Because the archive is a secondary copy, losing one Availability Zone does not lose the only copy, and One Zone-IA costs less than Standard-IA.
How Do You Detect Unsolicited Inbound Traffic in NAT Gateway Flow Logs?
Why is CloudTrail wrong for analyzing NAT gateway flow logs?
VPC Flow Logs are delivered to CloudWatch Logs (or S3/Firehose), not CloudTrail, which records AWS API activity rather than packet flows.
Why query dstAddr like 198.51.100.2 instead of the inbound record already seen?
The visible inbound record only shows the public IP sending traffic; the reverse flow with dstAddr 198.51.100.2 and srcAddr 203.0 shows whether the EC2 instance initiated the connection first.
How to Replicate Aurora MySQL Cross-Region for DR at Lowest Cost?
Why is DMS CDC to S3 not a valid Aurora DR solution here?
DMS writes change data as CSV or Parquet in S3, not a restorable Aurora cluster. Restoring would require ETL and reload, so it does not meet the database DR requirement.
Can Aurora automated backups be disabled to use hourly cross-Region backups?
Aurora automated backups cannot be disabled; retention is managed at the DB cluster level. That makes option D not implementable as written.
How to Collect Migration Evaluator Assessment and VM Dependencies with Least Overhead?
Why can't AWS Migration Hub alone generate the Quick Insights report?
Quick Insights is produced by Migration Evaluator; Migration Hub visualizes ADS discovery data but does not replace the Migration Evaluator Collector or its assessment workflow.
Is agentless Application Discovery enough for VM dependency mapping?
Agentless collection is useful for inventory, but application dependency visualization often needs agent-based inbound/outbound network connection data, especially when the scenario permits installing agents on every VM.
Choosing a Managed Directory and Admin Endpoint for EC2 Domain Join
Why can't Simple AD be used when MFA is required?
Simple AD is a Samba-based directory that lacks RADIUS/MFA integration and advanced policy features. AWS Managed Microsoft AD is required to integrate with RADIUS-based MFA providers for directory services.
Why is Amazon WorkSpaces preferred over an EC2 instance for administration here?
The question requires managed AWS services wherever possible. WorkSpaces is a managed desktop with MFA at sign-in, while an EC2 admin instance is self-managed and RADIUS MFA does not cover EC2 sign-ins.
When would an EC2-based admin host be the better choice?
When the directory supports many WorkSpaces — AWS advises against using a WorkSpace for administration at larger scale — or when custom tooling, domain controllers, or Group Policy management must be installed on a bastion host.
Decouple Product and Session Data for Cross-Region DR
Why is option B wrong if ElastiCache Global Datastore exists?
Global Datastore is supported only for Redis OSS 5.0.6 onward and Valkey 7.2. Amazon ElastiCache for Memcached has no cross-Region replication, so the DR Region would never receive a copy of the session data.
Can DynamoDB store structured product catalogue data?
Yes. DynamoDB handles structured and semi-structured data, and product catalogues are commonly modeled there, especially when access is key-based lookups rather than complex joins or ad-hoc SQL reporting.
How does DAX improve performance over a plain DynamoDB table?
DynamoDB Accelerator is an in-memory read cache that delivers microsecond-latency queries for hot items. Combined with global tables, it gives the fastest reads plus active-active multi-Region replication.
Fixing Stale EC2 Pricing Files with Mountpoint for Amazon S3
Doesn't Amazon S3 already guarantee strong consistency, so why is the pricing outdated?
S3 itself is strongly consistent; the problem is that each EC2 instance downloads the file once at launch and caches it locally for its entire lifetime. Later S3 updates never reach an already-running instance.
Why not use a Lambda function to copy prices into DynamoDB on each update?
It removes staleness but requires rewriting the ticketing service from flat-file parsing to per-item queries, and adds Lambda, DynamoDB read/write, and storage costs on every 1–15 minute refresh, so it is not the most cost-effective.
Is Mountpoint for Amazon S3 compatible with Linux EC2 instances?
Yes. Mountpoint for Amazon S3 is an open-source FUSE client for Linux, and it supports all S3 storage classes except the Glacier and Intelligent-Tiering archive tiers — S3 Standard works fine here.
Enforcing Cost Controls in AWS Control Tower Development OUs
Can I manually create SCPs for an OU managed by AWS Control Tower?
No. Manually creating or updating SCPs via AWS Organizations for an OU registered with Control Tower can cause the controls to enter an unknown state, requiring manual repair or re-registration of the OU.
What is the difference between a preventive and detective guardrail in Control Tower?
Preventive guardrails stop actions from occurring if they violate defined criteria (e.g., blocking non-burstable instances). Detective guardrails only monitor and report violations after resources are created.
How do Custom Guardrails work under the hood?
Custom Guardrails in AWS Control Tower automatically create and attach the necessary Service Control Policies (SCPs) to the target OU or account, ensuring compliance with the landing zone's governance model.
Cost-Effective Image Processing with SQS and S3 Lifecycle Policies
Why is SQS better than EventBridge for this scenario?
SQS provides a persistent message store with Dead-Letter Queues (DLQs), allowing you to inspect and manually or automatically rerun failed messages. EventBridge eventually drops events after retries are exhausted, making 'rerunning' harder without external state management.
Is moving images to Glacier Deep Archive cost-effective?
Yes, Glacier Deep Archive is the lowest-cost storage tier on AWS. Since the images are stored for up to 6 months and likely accessed infrequently after initial processing, this transition significantly reduces storage costs.
Lambda function URLs vs API Gateway for Git webhooks
Why is API Gateway HTTP API not the least-overhead choice here?
API Gateway is serverless and valid, but it adds an API layer, routes, stages, and integrations to configure. The scenario has no custom-domain, throttling, authorization, or centralized-routing requirement, so Lambda function URLs meet the need with fewer managed components.
When would option B become the better answer?
Option B wins if the company needs a single API front door, request throttling, WAF, custom domain names, or API-level authorization. Those requirements justify API Gateway's extra configuration.
Are Lambda function URLs a fully serverless endpoint?
Yes. They are managed HTTPS endpoints for Lambda functions, so no EC2, Auto Scaling group, or ALB is required, and Lambda scales the backing function automatically.
Cutting Cross-Region S3 Write Lag with Local Buckets and CRR
Why doesn't S3 Transfer Acceleration fix the lag from Lambda in sa-east-1 and ap-northeast-1?
Transfer Acceleration speeds up uploads that start on the public internet by entering AWS at a CloudFront edge location. Lambda functions already run on the AWS backbone, so there is no last mile to accelerate and the cross-Region distance remains.
Does using Cross-Region Replication break the requirement to centralize analysis in eu-north-1?
No. CRR replicates each local bucket's objects into the existing eu-north-1 bucket, so Athena continues to query one centralized dataset while ingestion happens close to each Region.
Why don't an S3 gateway endpoint or more Lambda memory reduce this lag?
A gateway endpoint only keeps VPC-to-S3 traffic private within a Region and does not shorten the inter-Region hop. Extra memory and multipart upload improve compute and large-object throughput, neither of which is the cross-Region bottleneck.
Direct Connect Backup: Why a Site-to-Site VPN Beats a Second VIF
Why can't MACsec be used on a 1 Gbps Direct Connect connection?
AWS supports MACsec encryption only on 10 Gbps and 100 Gbps Direct Connect ports. On the existing 1 Gbps link in this scenario, MACsec cannot be configured, so any option requiring it is invalid.
Do multiple private VIFs make a Direct Connect connection highly available?
No. Private VIFs are logical interfaces that share the same physical Direct Connect circuit, so a port or fiber failure takes down all of them simultaneously.
Why is a public Site-to-Site VPN a better backup for Direct Connect than a private-IP VPN?
A public Site-to-Site VPN travels over the internet on a separate physical path, so it survives a Direct Connect outage, while a private-IP VPN traverses Direct Connect itself and fails with it.
← Back to SAP-C02 AWS Certified Solutions Architect Professional Study Guide