SAP-C02
Deep Dive
Domain D4 (20%) tests your ability to select the right migration strategy for each application and the specific AWS tools to execute it. The 7 Rs, DMS, MGN and DataSync are guaranteed to appear on the exam.
Contents
The 7 Rs (formerly 6 Rs) is AWS's framework for classifying applications by migration strategy. Every application in the portfolio needs an assigned R based on its complexity, business value and dependencies.
Rehost
"Lift and Shift"Move the application to AWS with no changes. Same architecture, same configuration — only the underlying hardware changes.
Use when:
Fast migration, low risk, short timeline, no time for modernization.
Don't use when:
Doesn't take advantage of cloud benefits like auto-scaling or managed services.
Relocate
"Hypervisor Lift and Shift"Move on-premises VMware to VMware Cloud on AWS without converting to EC2. VMs keep their format.
Use when:
Heavy investment in on-premises VMware, want speed without recertifying equipment.
Don't use when:
Doesn't apply to apps on non-VMware hardware.
Replatform
"Lift, Tinker and Shift"Small optimizations without changing the core architecture. Move to managed services with minimal effort.
Use when:
Want managed service benefits without redesigning the app.
Don't use when:
Requires some adaptation work — not a pure lift.
Refactor
"Re-architect"Redesign the application to fully leverage cloud-native capabilities. More costly and slower, higher long-term value.
Use when:
App needs massive scale, lower operational cost, greater agility.
Don't use when:
Not viable for complex legacy apps with short timelines.
Repurchase
"Drop and Shop"Drop the on-premises software and replace it with a SaaS solution.
Use when:
Commodity software with no differentiated competitive advantage.
Don't use when:
If the app has deep customizations hard to replicate in SaaS.
Retire
"Decommission"Identify and remove applications that are no longer needed. Reduces portfolio cost and complexity.
Use when:
Typically 10–20% of the portfolio. Unused apps, duplicate functionality.
Don't use when:
If the app has active users or known dependencies.
Retain
"Revisit"Keep on-premises for now. For apps with compliance reasons, critical latency, or recent investment.
Use when:
Critical mainframes, apps with < 1ms latency, compliance prohibiting cloud.
Don't use when:
Should be temporary — revisit in the next migration cycle.
The CAF identifies organizational capabilities that must be developed for a successful migration. It organizes these capabilities into 6 Perspectives.
Business
Ensures IT is aligned with the business. ROI, business cases.
People
Organizational change management. Skills, roles, training.
Governance
Risk and compliance management. Portfolio management.
Platform
Systems and data architecture. Deploy processes.
Security
IAM, threat detection, incident response.
Operations
Monitoring, business continuity, operational change management.
MGN (formerly CloudEndure Migration) is AWS's rehost (lift-and-shift) service. It continuously replicates physical, virtual, or other-cloud servers into AWS with minimal cutover downtime.
MGN migration process
Install the Replication Agent
Agent on the source server that starts continuous block-level replication.
Continuous replication
Data is replicated in real time to a Staging Area in AWS (low-cost EC2).
Launch Test Instance
Launch a test instance to validate everything works without affecting production.
Cutover Window
Minimal cutover window (minutes). MGN converts the latest replica into the production instance.
MGN vs SMS (Server Migration Service)
AWS SMS — DEPRECATED
The old migration service. No longer receives new features. Exam questions mentioning SMS are being replaced by MGN.
MGN — Recommended
Continuous block-level replication (faster and more reliable), lower cutover downtime, supports more source platforms.
DMS migrates databases to AWS with minimal downtime using continuous change replication (CDC — Change Data Capture). It supports homogeneous migrations (Oracle→Oracle) and heterogeneous ones (Oracle→Aurora).
| Scenario | Tool |
|---|---|
| MySQL on-premises → Aurora MySQL | DMS alone (homogeneous — same engine family) |
| Oracle on-premises → Aurora PostgreSQL | AWS SCT (Schema Conversion Tool) + DMS |
| SQL Server → RDS SQL Server | DMS alone (homogeneous) |
| MongoDB → DynamoDB | DMS with a MongoDB source endpoint |
| Oracle/Teradata Data Warehouse → Redshift | AWS SCT + DMS |
| Continuous replication as CDC (ongoing replication) | DMS with ongoing replication tasks |
DMS Replication Instance
DMS work runs on a Replication Instance (managed EC2). Not on the source or the target. To minimize replication latency, deploy the replication instance in the same VPC/region as the target.
Current status: Snowcone was discontinued in November 2024, Snowmobile is retired, and Snowball Edge has not accepted new customers since November 7, 2025. For new customers, AWS recommends DataSync over Direct Connect (ideally a circuit dedicated to the migration) or AWS Data Transfer Terminal, a physical AWS facility where you bring your devices to upload at high speed; for edge compute, AWS Outposts. The table is kept for reference because the exam may still mention these devices.
| Device | Capacity | Edge compute? | Best for |
|---|---|---|---|
| Snowcone | 8–14 TB HDD/SSD | Yes (EC2, IoT Greengrass) | Discontinued (2024). Formerly: space-constrained edge, IoT, remote missions |
| Snowball Edge Storage Optimized | 80 TB per node | Yes (EC2, Lambda) | No new customers (Nov 2025). Formerly: NAS migration, datacenter backup, edge |
| Snowball Edge Compute Optimized | 42 TB + optional GPU | Yes (EC2, FPGA, GPU) | No new customers (Nov 2025). Formerly: edge ML, video in remote locations |
| Snowmobile | 100 PB per truck | No | Retired. Formerly: full datacenter migration (exabytes) |
When to use Snow vs DataSync/Direct Connect?
First calculate the network transfer time: 100 TB at 1 Gbps takes ~9 days (no overhead); at 100 Mbps, ~90 days. If that's too long, add dedicated capacity (a 10 Gbps Direct Connect moves 1 PB in ~9-10 days) and run DataSync in parallel, or bring the data to AWS Data Transfer Terminal. For recurring transfers once the network link is established → DataSync or Direct Connect.
AWS DataSync — accelerated data transfer
Transfers data from NFS, SMB, HDFS, on-premises S3 to S3, EFS, FSx. Up to 10× faster than standard tools. Automates integrity validation.
AWS Storage Gateway — hybrid extension
Connects on-premises applications to S3 storage transparently. Not for one-time migration — it's for ongoing hybrid integration.
Migration Hub and Application Discovery Service stopped accepting new customers on November 7, 2025. Their successor is AWS Transform, which uses AI agents to discover, plan, and track application migrations executed with MGN, DMS, and other tools.
Discovery and inventory
AWS Transform imports inventories (RVTools, CMDB, Migration Evaluator collector) with servers, dependencies and usage. Formerly: Application Discovery Service.
Waves and orchestration
Groups servers into waves and coordinates execution with MGN and DMS. Formerly: Migration Hub Orchestrator and Journeys.
Strategy recommendations
Analyzes the portfolio and code and recommends the strategy (the 7 Rs). Formerly: Migration Hub Strategy Recommendations.
Strangler Fig
Incremental monolith-to-microservices transition is done with API Gateway or an ALB routing by path (Refactor Spaces was part of Migration Hub).
Trap: "DMS can migrate any database with no additional tools"
Reality: For heterogeneous migrations (different engine, e.g. Oracle → PostgreSQL), you must use AWS Schema Conversion Tool (SCT) first to convert the schema. DMS handles the data migration, SCT converts the schema.
Trap: "Snowball Edge is still the default choice for moving petabytes"
Reality: Snowball Edge has not accepted new customers since November 2025. Today you use DataSync over a dedicated Direct Connect or AWS Data Transfer Terminal; for edge compute, AWS Outposts.
Trap: "MGN and SMS are equivalent and both in active use"
Reality: SMS is deprecated. MGN is the official replacement. If the exam mentions SMS in an answer option, treat it as a trap signal.
Trap: "Refactor is always the best migration strategy"
Reality: It depends on context. Refactor is the most expensive and slowest. For short timelines or apps about to be retired, Rehost is the right choice. The "best" strategy depends on business requirements.
Did you get this topic?
Test what you just learned
A company has 5 PB of data in an on-premises datacenter that must be migrated to Amazon S3 within a few weeks. They have a 1 Gbps internet connection to the datacenter. What is the most practical way to complete the migration on time?
Put what you learned to the test with original SAP-C02 exam questions.
Start SAP-C02 mock examUnlimited mock exams with the Pro plan
Log in to track your progress.