Determine a new architecture for existing workloads

Aldermere Council has containerized 25 ASP.NET applications that target .NET Framework 4.8, using Windows Server 2022 container images. The IT team has no Kubernetes experience and does not want to patch or scale container hosts. Each application runs 2 to 6 tasks behind an Application Load Balancer and must scale on CPU utilization.

Which container hosting platform should the solutions architect recommend?

1. AAmazon EKS with AWS Fargate profilesFargate on EKS runs Linux pods only, so it cannot run Windows containers, and the team also lacks Kubernetes skills.
2. BAmazon ECS on AWS Fargate with the Windows Server 2022 operating system family in the task definitionsECS on Fargate supports Windows containers, so the .NET Framework images run without any hosts to manage, and ECS service auto scaling and ALB integration cover the scaling and routing needs.
3. CAmazon ECS on Amazon EC2 Windows container instances in an Auto Scaling groupThis runs Windows containers, but the team would have to patch and scale the Windows hosts, which it explicitly wants to avoid.
4. DAmazon ECS Express Mode services running the Linux variants of the images.NET Framework 4.8 requires Windows, so Linux images would require porting the applications to modern .NET first.

Gilcrest Analytics runs a nightly risk calculation on a fixed cluster of 40 on-premises servers that sit idle 20 hours a day. The workload consists of about 2,000 independent jobs packaged as container images, each running 20 to 90 minutes, and jobs can be retried if interrupted. The company wants to migrate the workload, pay only while jobs run, minimize cost, and avoid building its own scheduler.

Which compute platform should the solutions architect choose?

1. AAWS Lambda functions invoked once per job from a container imageLambda supports container images, but a single invocation can run for at most 15 minutes, so jobs of 20 to 90 minutes cannot run on it.
2. BAWS Batch with a managed compute environment that uses EC2 Spot Instances and a job queueAWS Batch queues and schedules containerized jobs, scales compute from zero, and retries failed jobs. Because the jobs tolerate interruption, Spot capacity lowers cost substantially.
3. CA fleet of 40 EC2 On-Demand Instances covered by a Compute Savings PlanA fixed fleet recreates the idle capacity of the current cluster, and a Savings Plan commits to spend for hours when nothing runs. It also leaves the scheduling problem unsolved.
4. DAWS App Runner services, one per job typeApp Runner hosts request-driven web applications and APIs, not thousands of long, finite batch jobs, and it is no longer open to new customers.

Harrowgate Engineering stores 220 TB of CAD data on on-premises NetApp ONTAP arrays. Linux render nodes reach the data over NFS and Windows workstations reach the same volumes over SMB, and the storage team uses SnapMirror, snapshots, and thin provisioning with deduplication. The company wants a managed AWS service that keeps multiprotocol access and lets it replicate volumes to AWS with its existing ONTAP tools during the migration.

Which storage service should the solutions architect recommend?

1. AAmazon EFS with an AWS DataSync task from the NetApp arraysEFS supports NFS only, so the Windows SMB clients would lose access, and EFS cannot be a SnapMirror target.
2. BAmazon FSx for NetApp ONTAP, with SnapMirror replication from the on-premises arraysFSx for NetApp ONTAP runs ONTAP as a managed service, serves the same volumes over NFS and SMB, supports snapshots, deduplication, and thin provisioning, and can be a SnapMirror destination for migration.
3. CAmazon FSx for Windows File Server for SMB clients and Amazon FSx for Lustre for the render nodesSplitting the data across two file systems breaks shared access to the same volumes and gives up SnapMirror-based migration and ONTAP data management.
4. DAmazon S3 with Mountpoint for Amazon S3 on the render nodesS3 is object storage. Mountpoint provides file access for Linux reads and sequential writes, but it does not offer SMB, ONTAP features, or SnapMirror.

Wendmoor Utilities runs a vendor billing package on an on-premises Oracle Database. The vendor supports the package only if its monitoring agent is installed on the database host operating system and if certain database initialization parameters and OS packages are customized. Wendmoor wants AWS to handle automated backups, point-in-time recovery, and database provisioning wherever possible, and it does not want to manage Oracle on Amazon EC2 entirely by itself.

Which database platform should the solutions architect recommend?

1. AAmazon RDS for OracleRDS for Oracle is fully managed but does not allow access to the host operating system, so the vendor's agent and OS-level customizations cannot be installed.
2. BAmazon RDS Custom for OracleRDS Custom gives privileged access to the database host and operating system for vendor agents and customizations while still providing RDS automation such as automated backups and point-in-time recovery.
3. COracle Database self-managed on Amazon EC2EC2 allows full customization, but Wendmoor would have to build and run backups, recovery, and provisioning itself, which it wants to avoid.
4. DAmazon Aurora PostgreSQL-Compatible EditionThe vendor supports only Oracle, so converting the billing package to PostgreSQL would break the support agreement.