Prepare and test your skills
Prepare and test your skills
Free sample question, answered and explained below. The Generative AI Leader practice bank has 2,191 more like it. Try 10 free: instant score, no account.
An enterprise is building an architecture on Google Cloud to power real-time generative AI applications, including Retrieval-Augmented Generation (RAG) and foundational model integration. The data team needs to establish a storage and retrieval strategy that provides continuous uptime, rapid query latency, and seamless pipeline integration across Vertex AI workflows.
Which strategy best provides high data accessibility and efficient processing for generative AI workloads?
Cold archive storage (such as Archive class storage in Cloud Storage) is designed for infrequently accessed backup data and long-term regulatory compliance archives that are rarely read.
While cost optimization is an important architectural pillar, selecting cold archive storage for active data retrieval violates the fundamental AI requirement of data accessibility.
In Google Cloud, leveraging managed services such as Cloud Storage, BigQuery, and managed database systems like Cloud SQL (or Vertex AI Vector Search) provides a resilient foundation for enterprise AI. These services offer built-in high availability (HA), automated multi-zone replication, and seamless integration with Vertex AI pipelines.

This approach aligns with Google Cloud architecture best practices by decoupling compute from storage, eliminating single points of failure, and structuring data into consistent formats that downstream AI models and vector retrieval engines can query instantly.
This option relies on keeping operational data siloed in disparate, proprietary on-premises file formats and attempting to parse, transform, and convert data into readable text in real time whenever an end user submits an inference query.
Generative AI architectures rely heavily on pre-processed, standardized data formats stored in cloud environments to ensure minimal retrieval time during Retrieval-Augmented Generation (RAG).
This approach involves provisioning a single, unmanaged Compute Engine virtual machine and storing dataset files exclusively on its attached ephemeral or non-replicated local disk without external network backups or high-availability configurations.
Enterprise generative AI workloads require decoupled, resilient storage layers (such as managed Cloud Storage buckets and Cloud SQL/BigQuery instances) rather than isolated VM disks to ensure seamless integration and high availability.
Answer them yourself and get your score the moment you finish. No account needed.
Full access includes
Keep the momentum going with these hand-picked practice scenarios
Last updated