Task 4.1: Select existing workloads and processes for potential migration.

• ARehost, so that they move quickly to Amazon EC2 without changes

  Incorrect

  Rehosting moves applications unchanged, which would carry unused applications into AWS and keep paying to host them.

• BRetain, so that they stay in the data center until a later wave

  Incorrect

  Retain keeps applications in the source environment for now, for example because of compliance, risk or dependencies. It still keeps them running and costing money.

• CRetire, so that their servers are shut down and the data archived where needed

  Correct

  Retire is the strategy for applications with no business value; AWS guidance names applications with very low use and no inbound connections for 90 days as candidates.

• DRelocate, so that the servers move as they are to a cloud version of their platform

  Incorrect

  Relocating still moves and runs the applications, only on a different platform, so it keeps paying for something nobody uses.

Not every application should move. Unused applications found in the assessment are retired, which also shrinks the migration.

• AAWS Transform, using its discovery, dependency analysis and wave planning capabilities

  Correct

  AWS Transform collects and analyzes on-premises server inventory, groups applications, identifies dependencies and builds prioritized migration waves.

• BAWS Config with an aggregator for all accounts, recording the virtual machines as configuration items

  Incorrect

  AWS Config records the configuration of AWS resources. It does not discover virtual machines in a data center or plan migration waves.

• CAWS Compute Optimizer, opting in the data center hosts to receive sizing recommendations

  Incorrect

  Compute Optimizer analyzes AWS resources that already run in AWS accounts. It cannot inventory on-premises servers.

• DAWS Trusted Advisor, with its checks extended to the on-premises environment

  Incorrect

  Trusted Advisor inspects an AWS environment against best practices. It has no discovery or wave planning for on-premises servers.

Discovery, dependency mapping and wave planning for a new migration are capabilities of AWS Transform, which AWS now recommends in place of AWS Migration Hub and AWS Application Discovery Service, both of which no longer accept new customers.

• AEnter each server's current size into the AWS Pricing Calculator and total the monthly estimates

  Incorrect

  The Pricing Calculator estimates the cost of AWS services that you describe. Typing in today's sizes ignores actual utilization and does not collect data from the servers.

• BRequest an assessment with Migration Evaluator and collect utilization data with its agentless collector

  Correct

  Migration Evaluator builds a data-driven business case from discovered on-premises usage, recommends cost-effective AWS alternatives and compares bring-your-own-license with license-included options.

• CUse AWS Cost Explorer forecasting on a pilot account to project the cost of the whole server estate

  Incorrect

  Cost Explorer forecasts from spend that already happens on AWS. A pilot account says little about 900 servers that have not moved.

• DCreate AWS Budgets for the target accounts and compare those budgets with the current data center costs

  Incorrect

  Budgets track spend against a limit that you set. They do not estimate what the servers would cost on AWS.

A TCO business case built on real utilization and licensing options is what Migration Evaluator produces.

• AMove the order system in the first wave and the billing system in a later wave so that each wave stays small

  Incorrect

  Splitting systems that share a database leaves one of them talking to the database across the hybrid link between waves, which breaks the latency requirement.

• BOrder the waves by server operating system so that each wave uses one migration runbook

  Incorrect

  Grouping by operating system ignores the shared database dependency that AWS guidance asks wave planning to account for.

• CPut the order servers, the billing servers and the database cluster in the same move group and wave

  Correct

  Wave planning groups applications by their infrastructure and application dependencies, such as a shared database, so that servers that must stay together move together.

• DMove the database cluster first and leave both systems on premises until the last wave

  Incorrect

  Both systems would then reach the database across the hybrid link for most of the migration, which again breaks the latency requirement.

Dependencies decide move groups: servers that share a database and need low latency to it migrate in the same wave.

• ARefactor it into Lambda functions in a Region near the factory

  Incorrect

  Functions in a Region are still away from the factory floor.

• BRun it on AWS Outposts installed at the factory

  Correct

  Outposts extends AWS infrastructure, services and APIs to customer premises for lower latency and local data processing needs.

• CRehost it to EC2 in the nearest Region

  Incorrect

  A Region far from the factory adds network latency to the machines.

• DRetire the application

  Incorrect

  The application is needed for quality control.

Workloads that need local processing with AWS tools fit AWS Outposts.

• ASystems Manager Inventory in every account, synchronized to a central S3 bucket and queried for the installed database software

  Incorrect

  Inventory collects metadata from managed nodes into a central bucket, which shows where the software runs. It reports usage after the fact and cannot stop a launch that would exceed the licensed cores.

• BService Catalog portfolios that offer only approved database products, shared with every account that runs these workloads

  Incorrect

  A portfolio controls which products teams can launch, but it does not count cores against a license contract or stop the hundredth launch once the purchased cores are used up.

• CAWS License Manager with license configurations and hard limits, managed for all accounts of the organization

  Correct

  License Manager tracks licenses based on vCPUs, cores, sockets or machines across accounts and Regions of an organization. Administrators can set hard limits, so License Manager helps stop noncompliant usage before it happens.

• DThe AWS Config approved-amis-by-id managed rule, deployed to every account with the licensed database AMIs as its parameter

  Incorrect

  The rule marks running instances whose AMIs are not on the approved list as noncompliant. It detects after launch and knows nothing about how many licensed cores are in use.

The constraints are one view across accounts and Regions, counting by cores, and prevention rather than reporting. Inventory and a Config rule both detect after launch, and a Service Catalog portfolio controls what can be launched but not how much. License Manager with hard limits on license configurations counts cores organization-wide and stops launches that would exceed them.

• ARehost the feed handler and all 25 consumers together on Nitro-based instances in subnets of a transit gateway multicast domain

  Correct

  Rehosting moves the servers without changing the application. A transit gateway multicast domain delivers the feed between the subnets, and a sender must be a Nitro instance. Moving every consumer with the feed handler matters because multicast is not routed over Direct Connect.

• BRetain the application on premises, because VPC networking cannot deliver multicast traffic between EC2 instances

  Incorrect

  Retain fits applications that cannot or should not move yet, but the premise is wrong: a transit gateway multicast domain routes multicast traffic between subnets of attached VPCs.

• CRefactor the feed handler to publish prices to Amazon MSK topics that the consumers read with Kafka clients

  Incorrect

  Amazon MSK runs Apache Kafka, which suits a redesigned streaming platform, but moving producers and consumers to Kafka clients is a refactor of the code, which the company cannot do this year.

• DRehost the feed handler first and leave the consumers on premises to receive the multicast feed over Direct Connect

  Incorrect

  Transit gateway multicast routing is not supported over Direct Connect, Site-to-Site VPN, peering or Connect attachments, so consumers left on premises would stop receiving the feed.

An assessment should test the stated blocker before choosing Retain. AWS Transit Gateway supports multicast domains, so the application can be rehosted without code changes, but the details shape the plan: senders need Nitro instances, and because multicast does not cross Direct Connect or VPN, the feed handler and every consumer belong in the same move group.

• ARepurchase a SaaS product for every COBOL application

  Incorrect

  Custom banking applications rarely have SaaS equivalents for each one.

• BRetain the mainframe indefinitely

  Incorrect

  Leaders want to move away from the mainframe.

• CRehost the mainframe image to EC2 with AWS Transform MGN

  Incorrect

  Server replication rehosts virtual and bare metal servers; it does not run a z/OS mainframe image on EC2.

• DAWS Transform for mainframe modernization

  Correct

  AWS Transform can modernize and migrate IBM z/OS workloads to AWS.

Mainframe modernization on AWS is evaluated with AWS Transform.

• AA CloudHSM cluster in the workload's VPC as a custom key store

  Incorrect

  CloudHSM runs inside AWS, not in the insurer's own key manager.

• BAWS managed keys for each service

  Incorrect

  AWS managed keys are kept inside AWS KMS.

• CCustomer managed keys in KMS with automatic rotation turned on

  Incorrect

  Rotation changes key material but keeps it inside AWS KMS.

• DAn AWS KMS external key store backed by the insurer's key manager

  Correct

  External key stores protect AWS resources with keys outside of AWS, in an external key management system that you control.

Keys that must stay outside AWS are used through an external key store.