Hard Difficulty Questions

The instance's network bandwidth is saturated.

Why wrong: Network bandwidth saturation would show other symptoms like packet loss, not specifically high queue length.

The instance's CPU is under-provisioned.

Why wrong: CPU under-provisioning would show high CPU utilization, not necessarily high queue length.

The EBS volume is not encrypted.

Why wrong: Encryption does not affect queue length.

Use synchronous replication with a lower sync timeout setting to avoid blocking.

Why wrong: Lower timeout may cause frequent timeouts and data loss.

Keep synchronous replication but increase the replication timeout value in the HANA global.ini file.

Why wrong: Increasing timeout does not prevent performance impact; the source waits for acknowledgment.

Enable HANA delta merge operations during replication to reduce the amount of data transferred.

Why wrong: Delta merge is an internal operation, not a replication setting.

An EC2 Auto Scaling group configured for the SAP application tier

Why wrong: Auto Scaling groups are not required for LaMa; LaMa manages individual instances.

AWS Systems Manager managed instances for all SAP servers

Why wrong: Systems Manager is not a prerequisite; LaMa uses its own connector.

SSH keys for all SAP instances to allow LaMa to connect

Why wrong: LaMa uses AWS APIs, not SSH, to manage instances.

The route tables do not have a route for the SCS subnet.

Why wrong: Incorrect routing would cause persistent, not intermittent, failure.

The Internet Gateway is misconfigured.

Why wrong: Internet Gateway is not involved in private subnet communication.

The VPC peering connection is not established correctly.

Why wrong: No VPC peering is mentioned.

The EBS volumes are detached from the instance

Why wrong: No indication of volume detachment; the instance is running.

The instance state is 'running' so no issue with the instance

Why wrong: The instance can be running but the database may be down.

The security group does not allow inbound traffic to the database port

Why wrong: The command does not check security group rules.

Disable archiving of redo logs.

Why wrong: Disabling archiving is not safe and may cause data loss.

Switch from EBS gp3 to io2 Block Express volumes.

Why wrong: Improves I/O but not CPU.

Upgrade to a larger EC2 instance type with more vCPUs.

Why wrong: Does not reduce log write CPU usage.

Provision instance store volumes for HANA data and log areas.

Why wrong: Instance store is ephemeral; HANA requires persistent storage like EBS.

Purchase Reserved Instances for the HANA server to reduce costs.

Why wrong: Reserved Instances are a billing option, not a sizing requirement.

Select an RDS instance type that supports SAP HANA.

Why wrong: SAP HANA on AWS runs on EC2, not RDS.

The EBS volumes have reached the provisioned IOPS limit of 20,000 IOPS.

Why wrong: The volumes are striped; the aggregate IOPS is higher.

The S3 bucket is in a different region, causing network latency.

Why wrong: The bucket is in the same region.

The EC2 instance type does not match the requirements for Backint.

Why wrong: Backint does not require a specific instance type.

AWS Storage Gateway for on-premises caching

Why wrong: Storage Gateway is for hybrid, not directly for SAP cost optimization.

Amazon S3 Standard for all data

Why wrong: Standard is costly for infrequently accessed data.

A shared file system (e.g., Amazon EFS) for /sapmnt

Why wrong: A shared file system is important for SAP but not for session persistence.

A Multi-AZ deployment for the SAP application servers

Why wrong: Multi-AZ is for high availability of the database, not required for session persistence.

Configure HANA replication to Amazon S3 for point-in-time recovery.

Why wrong: S3 is for backup, not automatic failover.

Use HANA System Replication with a synchronous secondary in the same AZ.

Why wrong: Does not protect against AZ failure.

Deploy a single HANA database on a large EC2 instance in one AZ with regular backups to S3.

Why wrong: Single point of failure, no automatic failover.

Use an automation document to update the SSM Agent and schedule it via a cron job on each instance.

Why wrong: Manual scheduling on each instance is not scalable.

Enable the 'Auto-update SSM Agent' option in the EC2 launch settings for all instances.

Why wrong: This option is not available in launch settings; auto-update is a feature of the agent itself.

Configure AWS Systems Manager Patch Manager to update the SSM Agent.

Why wrong: Patch Manager updates OS patches, not the SSM Agent.

Detach the volume, create a snapshot, create a larger volume from the snapshot, and attach it

Why wrong: Detaching the volume causes downtime.

Create a new larger volume and copy data using rsync

Why wrong: Requires downtime for data copy.

Increase the IOPS of the volume to improve performance

Why wrong: Increasing IOPS does not increase storage capacity.

Increase the number of CPU cores on the database server.

Why wrong: CPU is already underutilized.

Enable Amazon EBS Elastic Volumes to increase volume size dynamically.

Why wrong: Volume size does not directly fix performance.

Disable backups on the source system to reduce load.

Why wrong: Backups are important for data protection.

Manually copy data files to AWS before starting replication.

Why wrong: HSR handles replication; manual copy is not needed.

The volume /dev/sdf has DeleteOnTermination set to false, which will cause data persistence issues.

Why wrong: False is the correct setting to prevent data loss.

The device names /dev/sdf and /dev/sdg are reserved for NVMe instance storage and should not be used for EBS.

Why wrong: These device names are valid for EBS.

The volumes are attached to a running instance, which prevents snapshot creation.

Why wrong: Snapshots can be taken of attached volumes.

i3.16xlarge

Why wrong: The i3.16xlarge has 488 GB of memory and is not SAP HANA-certified for large databases; it is intended for storage-intensive workloads, not large HANA deployments.

m5.24xlarge

Why wrong: The m5.24xlarge has 384 GB of memory, which is insufficient for a 3 TB HANA database and is not HANA-certified for this size.

r5.24xlarge

Why wrong: The r5.24xlarge has 768 GB of memory, which is not enough for a 3 TB HANA database, and it is not the recommended HANA-certified instance type for this workload.

NetworkIn

Why wrong: NetworkIn monitors incoming network traffic, which is not directly related to memory usage or query performance.

CPUUtilization

Why wrong: CPUUtilization measures CPU usage; while high CPU can be related, it does not directly indicate memory pressure.

DiskQueueDepth

Why wrong: DiskQueueDepth measures I/O queue depth, which is related to disk performance, not directly to memory issues.

CPUCreditBalance

Why wrong: CPUCreditBalance indicates CPU credit usage for burstable instances, not connection limits.

SwapUsage

Why wrong: SwapUsage indicates memory pressure, not connection limits.

ReadIOPS

Why wrong: ReadIOPS measures I/O operations, not connections.

Use an Application Load Balancer with sticky sessions enabled, and store sessions on each EC2 instance.

Why wrong: Incorrect. Storing sessions on each EC2 instance lacks a shared session store, so if an instance fails, sessions are lost. This does not provide high availability.

Use a Classic Load Balancer with sticky sessions and store sessions in RDS.

Why wrong: Incorrect. A Classic Load Balancer supports sticky sessions but lacks advanced features. Storing sessions in RDS can cause database bottlenecks and latency, not optimal for a distributed session store.

Use Amazon CloudFront with origin failover and store sessions in S3.

Why wrong: Incorrect. CloudFront is a content delivery network, not a load balancer. S3 is object storage with eventual consistency, not suitable for fast session storage.