Solutions Architect - Professional Practice Questions

AWS Direct Connect provides dedicated network connections from on-premises to AWS, offering consistent network performance and reduced bandwidth costs. Through the Direct Connect console, you can manage connections, view connection details including connection status and connection ID, accept hosted connections from AWS Direct Connect Partners, and delete connections when no longer needed. The connection ID is a unique identifier that looks similar to dxcon-fg5678gh and is used for API calls and support cases. Direct Connect bypasses the public internet for improved performance and security. Connection management capabilities include monitoring port status, viewing BGP session details, and configuring virtual interfaces for different VPCs or AWS services.

Yes, you can use AWS Direct Connect and VPN connections to the same VPC simultaneously, but primarily for failover scenarios. Direct Connect provides the primary path for consistent, high-bandwidth connectivity, while VPN serves as backup for redundancy. When both connections are active, Direct Connect will always be preferred for routing regardless of AS path prepending configurations. This hybrid approach ensures business continuity if the Direct Connect link fails. The VPN connection remains idle during normal operations but automatically becomes active if Direct Connect connectivity is lost. This architecture provides both high performance and reliability for critical workloads requiring consistent network connectivity.

AWS Data Pipeline is a web service that orchestrates and automates the movement and transformation of data between different AWS compute and storage services, as well as on-premises data sources, at scheduled intervals. Data Pipeline handles complex workflows involving data processing, transformation, and movement across multiple AWS services like S3, EMR, RDS, and Redshift. It provides built-in retry logic, failure handling, and monitoring capabilities to ensure reliable data operations. The service manages dependencies between data processing activities and can automatically provision and terminate resources as needed. Unlike simple data transfer tools, Data Pipeline provides workflow orchestration, scheduling, and error handling for complex data processing pipelines.

DynamoDB uses eventual consistency by default, meaning that after a write operation, it takes time for the update to propagate to all copies of the data across multiple Availability Zones. When you perform a read operation immediately after a write, you might not see the latest changes due to this propagation delay. Eventually consistent reads provide better performance and higher availability but may return stale data. For applications requiring immediate consistency, DynamoDB offers strongly consistent reads which guarantee that you receive the most up-to-date data, though at the cost of higher latency and lower availability during network issues.

A,C

A,C

Amazon Aurora Global Database provides the best disaster recovery solution for the specified RTO of 2 hours and RPO of 8 hours. Global Database enables cross-region replication with typical lag of less than 1 second, allowing for near-instantaneous failover to the secondary region when needed. The secondary cluster in us-west-2 can be promoted to primary within minutes, easily meeting the 2-hour RTO requirement. Multi-AZ deployments only protect against AZ failures within the same region, not regional disasters. EC2-based standby databases require manual failover and longer recovery times. Cross-region read replicas for RDS MySQL cannot be automatically promoted and would require manual intervention, making recovery times unpredictable.

DocumentDB with CloudFront, API Gateway, and Lambda provides the optimal solution for handling short but significant traffic spikes while maintaining fault tolerance across regions. CloudFront global edge locations automatically handle traffic spikes by caching responses and distributing load globally. Lambda automatically scales to handle concurrent requests without pre-provisioning capacity. API Gateway integrates seamlessly with Lambda for structured data operations. Aurora requires pre-provisioned capacity that may not handle sudden spikes efficiently. DynamoDB global tables would work but the combination lacks the global edge caching benefits of CloudFront. S3 is not suitable for structured data requiring read/write access patterns.

Amazon Simple Workflow (SWF) with G2 instances in a placement group provides the optimal solution for high-performance graphics rendering and video processing. G2 instances are GPU-optimized instances specifically designed for graphics-intensive workloads like rendering custom artwork and processing high-definition video content. SWF orchestrates complex workflows involving multiple steps of graphics processing, video encoding, and customer content integration. Placement groups ensure high network performance between instances for distributed rendering tasks. C3 instances lack GPU capabilities needed for graphics rendering. AWS Data Pipeline is designed for data movement and ETL, not real-time graphics processing workflows. The combination of SWF workflow orchestration with GPU-powered G2 instances provides the scale-out architecture needed for unpredictable bursts of graphics workloads while maintaining cost efficiency.

EC2-Classic addresses exhibit this behavior where Elastic IP addresses are disassociated when instances are stopped and must be manually reassociated upon restart. In EC2-Classic, stopping an instance releases the Elastic IP address back to the pool, requiring manual reattachment. This contrasts with EC2-VPC where Elastic IP addresses remain associated with stopped instances. VPC addresses maintain their association even when instances are stopped and restarted. The question specifically describes the unmapping and remapping behavior characteristic of EC2-Classic networking model, not VPC addressing.

Stateless instances with ElastiCache Memcached, Auto Scaling, and RDS read replicas provide the optimal architecture for a read-only news site with unpredictable traffic. Stateless instances enable rapid scaling since new instances do not need to maintain session state. ElastiCache Memcached provides high-performance caching for frequently accessed news content, reducing database load. RDS read replicas distribute read traffic across multiple database instances, essential for a read-heavy workload. Multi-AZ RDS provides high availability but does not improve read performance like read replicas do. Stateful instances are slower to scale and create bottlenecks during traffic spikes since they must maintain session information.