IBM Cloud Platform Application Development v2 Advanced Practice Exam: Hard Questions 2025

A development team has deployed a microservices application across multiple IBM Cloud Foundry regions for high availability. They notice that after a region failover, some user sessions are lost despite implementing session persistence. The application uses Cloud Foundry's sticky session routing, and session data is stored in Redis. Database transactions show no errors, but users report intermittent authentication failures during failover events. What is the MOST likely root cause and solution?

An enterprise application on IBM Cloud uses IBM Cloud Functions to process IoT events, with events stored in IBM Event Streams (Kafka). The application experiences significant performance degradation when processing spikes occur, despite having set appropriate concurrency limits. Monitoring shows that 30% of function invocations are timing out during peak loads, and Kafka consumer lag is growing exponentially. The team has already optimized function code and database queries. What architectural change would BEST address this scenario?

A financial services application on IBM Cloud requires strict data residency compliance, end-to-end encryption, and audit logging for all data access. The architecture uses IBM Cloud Object Storage for document storage, IBM Cloudant for metadata, and IBM App ID for authentication. During a compliance audit, it was discovered that some encrypted data written to Object Storage could potentially be accessed in decrypted form by IBM Cloud platform administrators. Which combination of changes would BEST ensure compliance with zero-knowledge encryption requirements?

A DevOps team manages a multi-tenant SaaS application on IBM Cloud using a continuous deployment pipeline with IBM Continuous Delivery. After implementing automated deployments, they experience a critical issue: approximately 5% of deployments result in database schema migration failures that leave some tenant databases in an inconsistent state while others complete successfully. Rollbacks are complex because each tenant has a separate database schema. The pipeline uses Tekton for orchestration and IBM Cloud Databases for PostgreSQL. What is the MOST robust solution to prevent partial migration failures?

An application on IBM Cloud Kubernetes Service experiences memory leaks that cause pods to be OOMKilled after 6-8 hours of operation. The team has implemented resource limits (memory: 2Gi limit, 1Gi request) and liveness probes, but the issue persists. Heap dumps show that the memory leak is caused by a third-party library that cannot be immediately replaced. The application must maintain 99.9% availability. Considering the constraint that the memory leak cannot be fixed immediately, what is the MOST effective mitigation strategy?

A globally distributed application uses IBM Cloud Internet Services (CIS) for DDoS protection and global load balancing across three regions. The application serves dynamic content that must be personalized per user. During load testing, the team discovers that 40% of requests are routed to suboptimal regions, increasing latency by 200-300ms. The CIS load balancer is configured with geographic steering, and all health checks pass. Origin servers in all regions are performing normally. What is the MOST likely cause and solution?

A development team uses IBM Cloud Continuous Delivery with a GitOps approach for managing Kubernetes deployments. They have implemented a pipeline that automatically applies changes when manifests are updated in the Git repository. After a recent deployment, the application became unavailable because a ConfigMap change was applied before the corresponding Deployment update, causing pods to crash due to missing configuration keys. The pipeline uses kubectl apply and runs in a single Tekton task. How should the pipeline be redesigned to prevent this ordering issue while maintaining GitOps principles?

An application uses IBM Cloud Functions to process images uploaded to IBM Cloud Object Storage. The workflow involves multiple chained functions: image validation, resizing, watermarking, and metadata extraction. During production, the team notices that approximately 15% of images fail processing with timeout errors, but only for images larger than 5MB. The individual functions complete successfully when tested in isolation. Each function has a 60-second timeout, and the entire sequence is orchestrated using action sequences. What architectural modification would MOST effectively resolve this issue?

A microservices application on IBM Cloud uses IBM Cloud Service Mesh (Istio) for service-to-service communication. The security team requires mutual TLS (mTLS) for all internal traffic and wants to implement fine-grained authorization policies. After enabling STRICT mTLS mode globally, several services begin experiencing intermittent connection failures with 'connection reset by peer' errors. The failures occur randomly and affect approximately 20% of requests to specific services. Istio sidecars are properly injected, and certificate rotation is functioning. What is the MOST likely cause and resolution?

A company runs a multi-region IBM Cloud Kubernetes Service deployment with IBM Cloud Databases for PostgreSQL configured with read replicas in each region. The application uses a write-through cache pattern with Redis. During a chaos engineering test simulating a regional failure, the team observed that while the application failed over successfully, data inconsistencies appeared where some users saw stale data for 10-15 minutes after failover. The PostgreSQL replication lag was under 2 seconds before failover, and the application correctly promoted a read replica to primary. What is the MOST likely cause of the data inconsistency?