D-CS-DS-23 Sample Questions & Answers

This is false. Factor XII (Admin processes) states that admin tasks should be run as one-off processes. This means they should be run in an identical environment as the regular long-running processes of the app, using the same codebase and configuration. Manually SSHing into a server can lead to inconsistencies and non-repeatable actions, which violates the principles of disposable and reproducible environments.

The Strangler Fig Pattern is specifically designed for incremental modernization of a legacy system. It involves gradually building new microservices around the old monolith, routing traffic to the new services as they become available. Over time, the new services 'strangle' the old system until it can be decommissioned. This approach minimizes risk and allows for a phased migration, which is ideal for complex, mission-critical applications.

RPO (Recovery Point Objective) refers to the maximum acceptable amount of data loss, measured in time. An RPO of 15 minutes means the business cannot lose more than 15 minutes of data. RTO (Recovery Time Objective) is the target time within which a business process must be restored after a disaster to avoid unacceptable consequences. An RTO of 1 hour means the service must be back online within one hour of the disaster.

The fundamental financial change when moving to the cloud is the shift from Capital Expenditure (CapEx) to Operational Expenditure (OpEx). Instead of purchasing physical servers and data center hardware upfront, organizations pay a recurring fee for the services they consume. This changes how IT is budgeted for, moving from a fixed asset model to a variable operational cost model.

Policy as Code (PaC) is the practice of defining and managing governance policies using code. Tools like Azure Policy, AWS Config Rules, or Open Policy Agent allow for the creation of rules that can automatically audit, enforce, or remediate non-compliant resources. This is the most effective and scalable way to enforce rules like mandatory tagging and instance type restrictions automatically, preventing non-compliant deployments before they happen.

A shift from project to product thinking involves moving away from temporary teams focused on delivering a fixed scope. Instead, long-lived 'product teams' are formed. These teams have end-to-end ownership of a service or product and are responsible for its development, operation, and continuous improvement. Their success is measured not just by delivery, but by the business outcomes and value the product generates over its entire lifecycle.

Managing user identities and permissions consistently across multiple clouds is a major challenge. A unified IAM solution (e.g., using a central IdP like Okta or Azure AD) is crucial for security and manageability.

Since data needs to move between the transcoding services in AWS and the analytics services in GCP, efficient inter-cloud networking is critical. This could involve direct interconnects or cloud exchange platforms to minimize latency and data transfer costs.

Data generated in one cloud (AWS) needs to be available in the other (GCP) for analytics. A robust data synchronization strategy is required to ensure data consistency and availability across the multi-cloud environment.

This option is the most comprehensive as it addresses all key aspects of the case study. It tackles the technical challenges with IaC for multi-cloud and CI/CD for agility. Crucially, it also addresses the workforce transformation by restructuring teams into cross-functional units and providing necessary training, which is essential for changing the culture and processes from waterfall to a collaborative, cloud-native model.

A service catalog is more than just a menu of options. For it to enable true self-service, it must be connected to a back-end automation and orchestration engine. When a developer requests a service (e.g., a pre-configured database or a Kubernetes namespace), this engine automatically executes the necessary IaC scripts and workflows to provision, configure, and deliver the resource without manual intervention, while enforcing all underlying governance policies.

The 'Concurrency' principle advocates for scaling out using the process model. Instead of making a single instance more powerful (scaling up or vertical scaling), the application's workload is distributed across numerous small, identical, and stateless processes (scaling out or horizontal scaling). This model is fundamental to cloud-native elasticity and resilience.