Question 580
A legacy order service will keep most application code but replace its commercial relational database with RDS for PostgreSQL. The candidate source engine uses dozens of proprietary stored procedures, and the migration team has not estimated which schema constructs can be converted. Before assigning the service to a low-risk replatform wave, what specific technical assessment is needed?
- Assume the database's data-export feature converts every proprietary stored procedure, so no source-code assessment is needed before cutover.
- Treat a successful source-engine backup as proof that RDS for PostgreSQL supports every source stored procedure and SQL dialect without application changes.
- Perform a target database schema and code compatibility assessment with DMS Schema Conversion or a comparable supported analysis, quantify manual conversion items and test the resulting application behavior before accepting the migration wave.
- Run a full-load AWS DMS test for table rows only and treat successful data transfer as proof that proprietary stored procedures will execute unchanged on PostgreSQL.
Explanation
Correct Answer
C:DMS Schema Conversion assesses schema/code conversion and highlights objects requiring manual intervention. Measuring and testing those gaps determines whether the proposed replatform is technically viable; this is distinct from merely naming the 7R strategy.
Incorrect Answers
A:Data movement and procedural code conversion are distinct concerns; moving table rows does not automatically translate incompatible stored procedures. The scenario explicitly identifies source-engine differences that must be assessed before asserting low risk.
B:Backup restorability protects data but does not make a heterogeneous database engine accept the previous engine's functions, triggers and procedural language. Runtime compatibility can still fail after the new database is populated.
D:A data migration can move rows while stored procedures and application SQL remain incompatible. DMS source-to-target data copying does not remove the need to assess and convert unsupported schema or database code objects.
Question 581
A retailer has approved a long-term Refactor strategy for a checkout monolith. Its cart service can be extracted first, but account and inventory functions will remain in the old monolith through several releases. The business needs an incremental rollout with a rapid fallback if the new cart service fails. Which migration control makes the intermediate architecture viable?
- Replace the entire monolith in one big-bang deployment and delete its old database before validating the new cart service.
- Keep the old and new cart implementations both accepting independent writes for months without a designated data owner or reconciliation mechanism.
- After extracting the cart service, perform a one-time public DNS cutover to the new cart endpoint for all callers without an interim proxy, interface adapter or tested rollback.
- Use the strangler-fig approach: route selected cart calls through a controlled proxy or adapter to the new service, maintain a clear source of truth and traffic rollback, then progressively retire replaced functionality.
Explanation
Correct Answer
D:Strangler-fig separates functionality incrementally while the monolith continues serving unmigrated routes. A proxy or anti-corruption adapter bridges the new and old interfaces, and controlled rollback protects customer transactions until the replacement proves safe.
Incorrect Answers
A:A one-step full rewrite forfeits the phased rollback explicitly required and exposes every checkout function to a simultaneous change. It is the opposite of an incremental extraction strategy for a high-value monolith.
B:Uncoordinated dual writes can create contradictory carts and inventory state even if both deployments remain reachable. Staged refactoring still needs clear ownership, synchronization or a controlled transition of writers, and rollback rules.
C:A one-off DNS switch can redirect some traffic but does not provide the selective internal interface bridging required while account and inventory code still calls the monolith. Nor does it give the controlled per-function fallback described in the scenario.
Question 582
A company runs 800 VMware VMs in a vCenter-managed on-premises estate. Its security team prohibits installing per-server discovery agents, and the migration team's immediate need is VM inventory, CPU/memory utilization and network-connection observations for capacity estimates—not process names inside each guest. Which AWS collection method is suitable?
- Install AWS Application Discovery Agents individually on all 800 VMs despite the stated security prohibition, since agentless inventory cannot collect any server utilization.
- Query AWS Config in the destination account to discover the on-premises vCenter VMs before those VMs are registered as cloud resources.
- Deploy the AWS Application Discovery Service Agentless Collector in the VMware environment with its supported modules and credentials to gather VM profiles and eligible utilization/network data without guest-agent installation.
- Use one vCenter VM configuration export showing provisioned RAM and CPU but no utilization history or network-connection collection, and treat allocation as representative workload demand.
Explanation
Correct Answer
C:Agentless Collector is deployed at the VMware environment rather than as a process on every guest VM, and its supported discovery modules collect the relevant inventory, utilization and network-traffic information. This tests a collector selection constraint, independently of another practice scenario's per-process physical-host agent requirement.
Incorrect Answers
A:The Discovery Agent can obtain richer per-process detail, but installing it conflicts with a hard organizational constraint. That extra detail is unnecessary for the explicitly scoped VMware host-level inventory and utilization request.
B:AWS Config is designed for supported AWS resource configuration history and compliance. It does not automatically discover VMs hidden inside a separate on-premises VMware environment without the specified discovery collection path.
D:Provisioned capacity alone does not reveal actual resource consumption or which VMs exchange data. Agentless Collector can gather the runtime utilization and network evidence required for migration sizing and dependencies, without installing guest agents.
Question 583
A planned migration wave includes stateful applications with a 30-minute approved interruption window. Teams intend to transfer 6 TB of data over a 100-Mbps link during the cutover itself. What feasibility check should stop this plan?
- Calculate achievable transfer and validation time, then design seeded replication and a smaller delta-based cutover if necessary.
- Increase AWS DataSync agent concurrency but assume a hard 100-Mbps source link will copy six terabytes during a thirty-minute freeze.
- Wait until cutover begins to discover the network route and assume replication will be instantaneous.
- Run an MGN migration test launch before transferring any source data and treat a boot check as proof a six-terabyte database has fully synchronized.
Explanation
Correct Answer
A:Network throughput sets a physical lower bound on bulk transfer, so pre-seeding is necessary when the cutover window is short.
Incorrect Answers
B:Protocol and task parallelism cannot exceed the source’s physical bandwidth ceiling; bulk transfer must be pre-seeded outside cutover.
C:Late discovery leaves no time to seed data or test destination synchronization.
D:A boot test does not calculate or eliminate the measured data-transfer time required for the selected stateful workload.
Question 584
A banking application has a zero-data-loss recovery target but the proposed migration architecture uses a once-per-day database snapshot in a secondary Region. The business owner insists the written requirements must not be relaxed. What planning action is needed?
- Approve the migration without disaster exercises and assume requirements are met once a backup file exists.
- Change the daily snapshot retention from one month to one year without changing capture frequency, and label the change zero-RPO protection.
- Hold the design decision and evaluate an architecture capable of the stated RPO with validated replication and failover behavior.
- Replicate the snapshot once each day to two separate Regions and assume geographic redundancy reduces the recovery point gap to zero.
Explanation
Correct Answer
C:A daily snapshot cannot provide zero data loss, so the migration architecture must be redesigned or formally re-scoped.
Incorrect Answers
A:A backup artifact does not prove the application can recover every committed transaction.
B:Longer retention preserves older recovery points but cannot eliminate daily intervals of unreplicated transactions.
D:Two stale replicas of a daily copy still omit transactions committed after the last successful snapshot.
Question 585
An organization uses AWS Organizations and has approved a multi-account landing zone. A migration pilot team wants to place regulated production services into an ungoverned sandbox because onboarding its production account takes longer. What must architecture require?
- Use the ungoverned sandbox for initial regulated workloads and promise to attach the production OU control baseline after customer data is already loaded.
- Use a single public VPC subnet for all services and assume public internet access replaces account guardrails.
- Provision the intended governed account and OU with logging, security and identity controls before allowing production data migration.
- Enable only AWS Budgets alerts in the sandbox and assume spending notifications substitute for account guardrails, log archival, and production identity controls.
Explanation
Correct Answer
C:Production readiness includes account boundaries and mandatory controls, not merely availability of EC2 capacity.
Incorrect Answers
A:The production security and compliance guardrails are prerequisite authorization boundaries, not optional post-cutover cosmetic settings.
B:Network exposure cannot substitute for required organizational policy and logging foundations.
D:Budgets cost controls do not create the governance and audit boundaries required for sensitive production workloads.
Question 586
A regional logistics application relies on a third-party connectivity appliance that is unsupported on EC2, but its operator can offer the same functionality through a managed AWS Marketplace appliance with a supported contract. Which pre-migration choice needs evaluation?
- Select unsupported direct rehost on EC2 because the source appliance boots, while ignoring the vendor’s explicit supported managed replacement offering.
- Purchase the supported Marketplace appliance and immediately switch dependencies without validating protocol compatibility, licensing entitlement or network placement.
- Compare a supported repurchase or substitution path with rehost feasibility, including licensing and integration validation.
- Assume the legacy binary will become vendor-supported merely because its virtual disk boots on EC2.
Explanation
Correct Answer
C:Product support and procurement constraints can make a supported alternative safer than attempting an unsupported lift-and-shift.
Incorrect Answers
A:Boot compatibility is not contractual or feature-support evidence for an appliance whose vendor does not certify EC2 rehosting.
B:A managed alternative may be viable but still requires contract, functional and integration validation before cutover.
D:Technical boot success does not establish vendor support, license compliance or reliable operation.
Question 587
During application assessment a security team discovers an application contains a hard-coded database admin password and has no auditable service identity. Its functional owner wants to move it unchanged into the production OU next week. Which migration readiness decision is appropriate?
- Use EC2 user data to inject the same plaintext admin password into every migrated host without adding scoped service identity or secret rotation.
- Require remediation of the credential and access model, or a formally approved risk exception with compensating controls, before production cutover.
- Copy the same plaintext database password into every EC2 launch script and declare the migration secure.
- Grant the migrated application a shared IAM administrator role so database users no longer need individual attribution.
Explanation
Correct Answer
B:Known privileged credential exposure is a readiness risk that does not disappear when the hosting location changes.
Incorrect Answers
A:Relocating a hardcoded high-privilege credential into user data retains the secret exposure and auditability flaw.
C:Replicating a privileged static secret broadens exposure and preserves the defect.
D:A broad shared IAM role can widen the exposure and does not provide least-privilege auditable database credential use.
Question 588
A portfolio inventory includes a database with strict data-residency requirements but an analytics component may lawfully process only aggregated anonymized output in another Region. The proposed wave treats both datasets as identical. What classification work is necessary?
- Separate datasets by sensitivity and residency, document approved processing paths and design regional boundaries for each data class.
- Send both raw and anonymized datasets to the analytics Region solely because aggregated output is legally allowed there.
- Assume anonymization is automatic whenever a CSV file is copied to a second Region.
- Use a generic organization-wide S3 replication rule for both datasets without separate data-residency and reidentification controls.
Explanation
Correct Answer
A:Data classification may allow different placement and transfer conditions for raw regulated data and lawful derived aggregates.
Incorrect Answers
B:The permission for derived anonymized data does not authorize exporting the raw regulated dataset.
C:File format and location do not prove that personal information was irreversibly anonymized.
D:Broad replication does not enforce different regional boundaries for protected raw data versus approved derivatives.
Question 589
A migration portfolio contains a service that requires 99.99% availability, but its only current dependency is a local licensing server with no remote failover. The architects want to set a cutover date based solely on EC2 launch readiness. What must be addressed first?
- Add more application servers across AZs but continue using the one on-premises licensing server and untested single WAN route.
- Identify the licensing server as a critical dependency and design/test an available supported access path during and after migration.
- Move the web layer first and assume the remote licensing service will always be reachable after cutover.
- Move the licensing server to a different unsupported EC2 image without vendor approval and assume the required license semantics will be preserved.
Explanation
Correct Answer
B:Migrating a compute tier does not remove a single-point licensing dependency that can prevent application operation.
Incorrect Answers
A:Front-end compute redundancy cannot overcome a mandatory external license service that remains a single point of failure.
C:An unverified licensing path can break an otherwise healthy migrated application.
D:License and service availability need validated vendor-supported deployment and tested access from the migrated application.
Question 590
A manufacturing organization plans a cloud migration wave near a legally mandated audit freeze. Operations cannot change software during the freeze, but the migration coordinator has scheduled a database engine upgrade and infrastructure cutover on the same date. Which sequencing decision reduces regulatory and rollback risk?
- Treat database engine upgrade and infrastructure cutover as one irreversible change during the audit freeze, without testing rollback separately.
- Separate unsupported simultaneous changes and schedule pilot, upgrade and cutover windows around the approved freeze with clear rollback gates.
- Perform all changes during the freeze without a documented exception because cloud migration dates override controls.
- Perform an out-of-window upgrade in production and rely on S3 backups without testing database-level compatibility or restoring data.
Explanation
Correct Answer
B:Distinct change stages reduce compounded failure causes and preserve audit-approved change control boundaries.
Incorrect Answers
A:Combining unapproved simultaneous changes defeats the mandated freeze and obscures which change caused a failure.
C:Ignoring a mandated freeze creates compliance risk and removes planned rollback capacity.
D:Backups alone do not authorize a change during a freeze or demonstrate that the upgrade and cutover can be rolled back safely.
AWS Certified Solutions Architect - Professional SAP-C02 Practice Tests