I85PREP
⌘ K
IBM Cloud Pak for Data v4.7 Architect study materials
IBM Cloud Pak for Data v4.7 Architect study resources

IBM Cloud Pak for Data v4.7 Architect practice questions

One question at a time. Build your IBM Cloud Pak for Data v4.7 Architect understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Data and AI architecture

On IBM Cloud Pak for Data v4.7, which control plane component is the central entry point that brokers authentication, tenant isolation, and access to platform services such as catalogs and projects?

  1. The Watson Studio runtime pods
  2. The Cloud Pak for Data CLI
  3. The Db2 Warehouse head node
  4. The Zen control plane services Correct answer.

Explanation

The Zen services form the platform control plane: they host the web client, handle identity and token exchange, enforce tenant and role-based access, and route requests to service APIs. Runtime pods and database nodes are workload components that sit behind this control plane.

Question 2Data and AI architecture

A team wants to give business analysts a governed, curated collection of assets drawn from multiple catalogs without duplicating the underlying data. Which Cloud Pak for Data capability should the architect recommend?

  1. A new physical copy of the data in a governed schema
  2. A metadata import enrichment run on each catalog
  3. A data virtualization layer exposed through a virtualized project Correct answer.
  4. A separate Cloud Pak for Data instance per business unit

Explanation

Virtualized assets let consumers query data in place across sources through a single governed access layer, avoiding physical duplication and its synchronization cost. Catalog enrichment affects metadata only, and copying data or standing up separate instances increases cost and fragments governance.

Question 3Data and AI architecture

Which statement best describes the role of projects versus catalogs in Cloud Pak for Data v4.7?

  1. Projects are collaborative workspaces for analysis and assets, while catalogs govern and share assets across the organization Correct answer.
  2. Projects and catalogs are interchangeable containers that differ only in their display icon
  3. Catalogs are collaborative workspaces, while projects are read-only inventories of governed assets
  4. Projects store only notebooks, while catalogs store only database connections

Explanation

Projects provide an isolated collaborative workspace where users add assets, run analyses, and promote results. Catalogs are the governance and discovery layer where curated assets are published, enriched, and shared with broader audiences under policy.

Question 4Data and AI architecture

An architect needs row-level and column-level access control enforced consistently for users querying governed data through multiple engines. Which approach fits Cloud Pak for Data governance best?

  1. Encrypt each column with separate keys distributed to each business unit
  2. Grant database roles directly on each engine and document the mappings in a spreadsheet
  3. Define policies once in the governance layer and let integrated engines enforce them at query time Correct answer.
  4. Build per-team views in every engine and rely on users to select the right view

Explanation

Central policy definition with enforcement pushed to integrated engines keeps access rules consistent and auditable across tools. Engine-by-engine grants and hand-built views drift over time and are hard to prove compliant, and per-unit encryption keys do not express row or column policy.

Question 5Data and AI architecture

Which Cloud Pak for Data service is the metadata repository that underpins catalog search, lineage, and governance artifacts?

  1. DataStage
  2. Analytics Engine powered by Apache Spark
  3. Watson Knowledge Catalog Correct answer.
  4. Watson OpenScale

Explanation

Watson Knowledge Catalog maintains the metadata repository and governance artifacts such as catalogs, glossary terms, policies, and lineage that other services reference. The other options are model monitoring, data integration, and compute services that consume rather than own this metadata.

Question 6Data and AI architecture

A client requires that analytic workloads run in an isolated network zone while still using the shared platform identity provider. Which design decision is most appropriate?

  1. Disable authentication in the isolated zone and rely on network segmentation alone
  2. Merge the isolated workloads into the main namespace and use firewall rules only
  3. Deploy a second identity provider inside the isolated zone and synchronize users manually
  4. Deploy the workloads in a separate namespace or instance that trusts the central identity provider for authentication Correct answer.

Explanation

Namespace or instance separation provides workload and network isolation while federating authentication to the central identity provider keeps a single identity source. A second identity provider or no authentication at all breaks the single sign-on and audit model.

Question 7Data and AI architecture

What is the primary purpose of a Cloud Pak for Data service instance (for example, an instance of Watson Studio)?

  1. To define the physical network topology of the cluster
  2. To compile notebooks into container images for production
  3. To provision an isolated deployment of a service with its own configuration, storage, and access controls Correct answer.
  4. To register a user account in the platform identity provider

Explanation

A service instance is a provisioned, configured deployment of a service that customers create within a namespace, with its own settings, storage, and access boundaries. Identity registration, network topology, and image builds are handled by other platform mechanisms.

Question 8Data and AI architecture

An architect is designing a data science workflow that must train models on sensitive records while keeping raw values out of the analyst workspace. Which combination best meets the requirement?

  1. Grant analysts direct SELECT on the source tables and rely on policy review
  2. Use data masking or substitution in a governed virtualized view and train on the masked projection Correct answer.
  3. Copy the raw records into each analyst's project for convenience
  4. Export the raw records to a flat file stored in the analyst's home directory

Explanation

Masked or substituted projections delivered through a governed virtualization layer let analysts train on realistic structure without exposing raw sensitive values. Copying or exporting raw data and granting direct access all bypass the protection the requirement demands.

Question 9Data and AI architecture

Which statement about Cloud Pak for Data deployment on Red Hat OpenShift is accurate for v4.7?

  1. The platform requires installation directly on bare-metal nodes without a container platform
  2. The platform runs as containerized operators and custom resources managed by OpenShift Correct answer.
  3. The platform runs only on a single node because it cannot scale horizontally
  4. The platform replaces OpenShift with its own container orchestrator

Explanation

Cloud Pak for Data is delivered as operators and custom resources that OpenShift schedules and manages, which is how services are installed, upgraded, and scaled. It depends on OpenShift rather than replacing it, and it scales across multiple worker nodes.

Question 10Data and AI architecture

A customer wants a single governed query endpoint over Db2, object storage, and a SaaS database, with pushdown where possible. Which architectural component is central to this design?

  1. A batch scheduler that copies data nightly into one warehouse
  2. A data virtualization service that federates queries to remote sources Correct answer.
  3. A message queue that buffers query results between systems
  4. A block storage volume mounted into every project

Explanation

A virtualization or federation service presents one governed SQL endpoint and pushes processing to remote sources where the connector supports it. Queues, shared block volumes, and nightly copies do not provide federated query access.

Question 11Data and AI architecture

Which capability in Watson Knowledge Catalog automatically discovers and classifies sensitive data such as personal identifiers across connected sources?

  1. Manual tagging performed by every data steward after each load
  2. Automated discovery and sensitive-data classification through analysis of sampled data and metadata Correct answer.
  3. Query history reports exported from the database engine
  4. Row-count reconciliation reports generated by the ETL job

Explanation

Automated discovery profiles sampled data and metadata to infer asset types and assign classifications such as personal information, which then feed governance policies. Manual tagging, row counts, and query history do not classify sensitive content at scale.

Question 12Data and AI architecture

An architect must design lineage reporting so that a regulator can trace a reported metric back to its source columns. Which practice is most important?

  1. Capture lineage across ingestion, transformation, and consumption and retain it as governed metadata Correct answer.
  2. Rely on developers to document transformations in code comments
  3. Store transformation logic in emails sent to the data steward
  4. Record only the final dashboard definition and its owner

Explanation

End-to-end lineage captured as governed metadata links reported metrics to source columns through every transformation step and survives staff changes. Dashboard-only records, code comments, and emails are incomplete and not auditable.

Question 13Data and AI architecture

Which statement best characterizes the relationship between assets in a project and assets published to a catalog?

  1. Publishing moves the asset so it no longer exists in the project
  2. Catalog assets cannot be used inside projects once published
  3. Project assets are automatically published to every catalog on creation
  4. Publishing promotes a project asset into a catalog where it gains governance, enrichment, and broader discoverability Correct answer.

Explanation

Publishing creates a governed catalog entry that can be enriched, rated, and shared while the project asset remains available for continued work. It is not a destructive move, it is not automatic to all catalogs, and published assets can be added to projects.

Question 14Data and AI architecture

A client needs to run Spark jobs on a managed cluster without operating Hadoop infrastructure themselves. Which Cloud Pak for Data service addresses this?

  1. Watson Knowledge Catalog
  2. Analytics Engine powered by Apache Spark Correct answer.
  3. Cloud Pak for Data control plane
  4. Watson OpenScale

Explanation

Analytics Engine provides managed Spark compute for notebooks, jobs, and SQL workloads without the customer operating the cluster infrastructure. The other options address model monitoring, metadata governance, and platform control respectively.

Question 15Data and AI architecture

An architect is asked to reduce risk when analysts use generative AI assistants over governed enterprise data. Which design choice best aligns with platform governance?

  1. Disable audit logging to reduce overhead during pilot use
  2. Copy sensitive datasets into the assistant's prompt history for reuse
  3. Allow the assistant to query source databases with a shared superuser credential
  4. Ground prompts in governed catalog assets and enforce the same access policies on retrieval Correct answer.

Explanation

Grounding retrieval in governed assets and reusing platform access policies keeps the assistant within the same authorization and audit boundaries as other consumers. Shared superuser credentials, prompt-history copies, and disabled logging all undermine governance.

Question 16Data and AI architecture

Which component is responsible for storing and serving the platform's user interface and API gateway for Cloud Pak for Data clients?

  1. The Db2 backup utility
  2. The Spark runtime images
  3. The object storage bucket used for notebooks
  4. The Zen control plane services Correct answer.

Explanation

Zen services host the web client and API gateway that clients reach, and they route requests to the appropriate service APIs after authentication. Spark images, backup utilities, and storage buckets are supporting components with narrower roles.

Question 17Data and AI architecture

A data steward needs to define business terms and attach them to physical assets so analysts see consistent definitions. Which governance artifact should be used?

  1. A project notebook that documents column names
  2. A database comment on each physical column
  3. A dashboard that lists column descriptions
  4. A business glossary with terms linked to catalog assets Correct answer.

Explanation

A business glossary holds governed business terms that can be linked to technical assets, giving analysts consistent definitions across sources. Notebooks, column comments, and dashboards may describe data but are not the governed glossary artifact.

Question 18Data and AI architecture

An architect must ensure that a newly connected cloud object storage source can be profiled and governed without moving the files. Which sequence is most appropriate?

  1. Create a connection, discover and profile the assets in place, then publish curated assets to a catalog Correct answer.
  2. Grant every user direct bucket credentials and skip catalog registration
  3. Create a duplicate bucket inside the platform and copy all objects into it
  4. Download all objects into a project, then delete the source connection

Explanation

Connection-based discovery and profiling read metadata and samples in place, after which curated assets are published for governed consumption. Downloading, duplicating, or bypassing the catalog all add cost or remove governance without being necessary.

Question 19Data and AI architecture

Which statement about namespaces in a Cloud Pak for Data deployment is correct?

  1. Only one namespace may exist per cluster and it must be named default
  2. Namespaces are created automatically per user and cannot be shared
  3. Namespaces control only network routes and never hold service instances
  4. A namespace scopes service instances and their resources, and multiple namespaces can exist on one cluster Correct answer.

Explanation

OpenShift namespaces provide scoping for service instances, secrets, and related resources, and a cluster can host several of them to separate teams or environments. They are not limited to one per cluster, not per-user by default, and they do hold service instances.

Question 20Data and AI architecture

A customer wants to compare model predictions against actual outcomes over time and detect drift in production. Which service should the architect include?

  1. Analytics Engine powered by Apache Spark
  2. DataStage
  3. Watson OpenScale Correct answer.
  4. Watson Knowledge Catalog

Explanation

Watson OpenScale evaluates deployed models against payload and feedback data to measure quality, fairness, and drift over time. DataStage integrates data, Analytics Engine provides compute, and the catalog governs metadata; none performs production model monitoring.

Question 21Data and AI architecture

An architect is choosing between a physical warehouse and a virtualized access layer for a reporting workload with strict freshness requirements across many sources. Which consideration most strongly favors virtualization?

  1. Reports need near-real-time values from sources that cannot be replicated on a short cycle Correct answer.
  2. The team wants to minimize query cost by pre-aggregating all data
  3. Reports run once a month and can tolerate day-old data
  4. The sources are already consolidated in a single high-performance warehouse

Explanation

Virtualization avoids replication latency, so it suits workloads that need current values from sources that are impractical to copy frequently. Monthly reporting, pre-aggregation for cost, and already-consolidated sources are cases where physical or pre-aggregated approaches are usually preferable.

Question 22Data and AI architecture

Which practice best supports collaboration when several data scientists must share notebooks, connections, and results inside Cloud Pak for Data?

  1. Share a single administrator credential among the whole team
  2. Create one personal project per person and copy assets manually
  3. Email notebook files to each other and run them locally
  4. Work in a shared project with role-based membership and promote finalized assets to a catalog Correct answer.

Explanation

A shared project provides role-based membership, common connections, and shared assets so collaborators work on the same state. Emailing files, manual copying between personal projects, and shared credentials all break collaboration controls and auditability.

Question 23Data and AI architecture

An architect must enforce that only approved users can view a catalog asset containing personal data, while other catalog users may still see that the asset exists. Which mechanism fits best?

  1. Apply governance policies and membership rules that restrict asset access while keeping metadata discoverable Correct answer.
  2. Encrypt the asset name so it cannot be searched
  3. Remove the asset from the catalog and distribute its location by email
  4. Grant all catalog users full access and rely on an honor system

Explanation

Governance policies and catalog membership controls can restrict who opens an asset while leaving its metadata visible for discovery and request workflows. Removing the asset, hiding its name, or granting open access either destroys discoverability or removes the control entirely.

Question 24Data and AI architecture

Which statement about the Cloud Pak for Data CLI and REST APIs is correct for automation design?

  1. They replace OpenShift and manage cluster nodes directly
  2. They can only read data from Db2 and cannot change platform configuration
  3. They are limited to interactive browser sessions and cannot be automated
  4. They expose platform operations so provisioning and governance tasks can be scripted against the control plane Correct answer.

Explanation

The CLI and REST APIs wrap control plane operations, enabling scripted provisioning, catalog management, and governance actions in pipelines. They do not manage cluster nodes directly, are not read-only, and are designed for automation rather than only interactive use.

Question 25Data and AI architecture

A regulated client requires that data remain in a specific country while a global team analyzes it through the platform. Which architecture satisfies the constraint?

  1. Let analysts download extracts and analyze them locally abroad
  2. Replicate all data to a central region and delete the regional copy
  3. Deploy the data-holding platform instance in the required region and give the global team governed remote access Correct answer.
  4. Store the data in a global bucket without regional controls

Explanation

Keeping the instance and storage in the required region satisfies residency while governed remote access supports the global team. Central replication, local extracts abroad, and uncontrolled global buckets all move or expose data outside the mandated location.

Question 26Data and AI architecture

An architect is planning a Cloud Pak for Data v4.7 installation and wants to understand how the platform's core services are packaged for Red Hat OpenShift. Which statement describes the platform's deployment model?

  1. It deploys only to virtual machines and cannot run inside OpenShift namespaces.
  2. It installs as a single monolithic application container that bundles every service into one pod.
  3. It installs as a set of operators and custom resources managed through the platform's own control plane on OpenShift. Correct answer.
  4. It requires a dedicated non-Kubernetes cluster manager installed before OpenShift.

Explanation

Cloud Pak for Data v4.7 is delivered as operators and custom resources that run on Red Hat OpenShift, with the platform control plane coordinating service installation and lifecycle. This operator-based model lets architects install only the services they need and manage them declaratively.

Question 27Data and AI architecture

A team wants analysts to query data in a remote Db2 warehouse without copying it into the platform's own storage. Which Cloud Pak for Data capability supports this?

  1. Creating a connection and adding the remote table as a virtualized data asset. Correct answer.
  2. Replicating the table nightly into the platform's internal metadata store.
  3. Registering the table only as a business term in a glossary.
  4. Exporting the remote table to a CSV file and uploading it into a project.

Explanation

Virtualized assets use a connection to the remote source and expose its tables for query without physically moving the data. This is the standard approach when an architect must avoid duplication and keep data at its source.

Question 28Data and AI architecture

Which Cloud Pak for Data component provides the shared metadata repository that catalogs assets, governs them, and enforces access policies?

  1. Watson Studio
  2. Watson Machine Learning
  3. DataStage
  4. Watson Knowledge Catalog Correct answer.

Explanation

Watson Knowledge Catalog is the governance and catalog layer that stores asset metadata, business terms, and policies. It is the component architects use to make assets discoverable and to enforce governance across the platform.

Question 29Data and AI architecture

An architect must explain how projects and catalogs differ in Cloud Pak for Data. Which statement is accurate?

  1. A project is a governed collection for the whole enterprise, while a catalog is a private scratch area for one user.
  2. A catalog can contain only notebooks, while a project can contain only data assets.
  3. Projects and catalogs are the same object with different labels in the interface.
  4. A project is a collaborative workspace for building and analyzing assets, while a catalog is a governed, curated collection for discovery by a broader audience. Correct answer.

Explanation

Projects are working spaces where team members create and refine assets, and catalogs are curated, governed collections that publish selected assets to a wider audience. Promoting an asset from a project to a catalog is how work becomes discoverable and governed.

Question 30Data and AI architecture

A client wants to run Spark workloads on a managed Spark cluster inside Cloud Pak for Data rather than operating Hadoop themselves. Which service should the architect select?

  1. Analytics Engine powered by Apache Spark Correct answer.
  2. Watson Discovery
  3. Watson OpenScale
  4. Db2 Warehouse

Explanation

Analytics Engine powered by Apache Spark provides managed Spark runtime and clusters within the platform, so teams can run Spark jobs without building Hadoop infrastructure. The architect provisions Spark instances through this service.

Question 31Data and AI architecture

An architect is asked where identity and access are enforced when users sign in to Cloud Pak for Data. Which statement is correct?

  1. Users are authenticated solely by the database connection credentials of each data source.
  2. Identity is enforced only at the browser through client-side checks.
  3. Each service maintains its own separate user directory with no shared identity.
  4. Authentication is delegated to an enterprise identity provider through OpenShift and platform configuration, and authorization is enforced per service and asset. Correct answer.

Explanation

The platform integrates with enterprise identity providers so users authenticate once, and authorization is then applied by platform and service-level policies on assets and resources. Client-side checks alone cannot provide security.

Question 32Data and AI architecture

A data steward needs to attach business definitions to physical columns so analysts see consistent meaning. Which Cloud Pak for Data construct should the architect recommend?

  1. Connection credentials stored in a shared project.
  2. A scheduled job that renames every column in the source system.
  3. Business terms in a governed glossary, assigned to the physical assets in the catalog. Correct answer.
  4. Notebook markdown cells that describe the columns.

Explanation

A governed glossary holds business terms, and stewards assign those terms to cataloged physical assets so the same definition appears wherever the asset is used. This preserves consistent meaning without altering source systems.

Question 33Data and AI architecture

An architect must let analysts query data from several sources through one governed SQL endpoint with pushdown where possible. Which Cloud Pak for Data capability fits?

  1. A virtualized data layer built on connections to each source, queried through the platform's SQL engine. Correct answer.
  2. A model deployment endpoint that scores each row on request.
  3. A glossary that lists each source table as a business term.
  4. A separate physical copy of every source loaded into one project.

Explanation

Virtualization exposes remote tables through connections and lets the platform's query engine push work to sources where supported. This gives a single governed query surface without duplicating data.

Question 34Data and AI architecture

A client needs to detect drift in a deployed model by comparing predictions with actual outcomes over time. Which Cloud Pak for Data service should the architect propose?

  1. Watson Knowledge Catalog
  2. DataStage
  3. Watson OpenScale Correct answer.
  4. Analytics Engine powered by Apache Spark

Explanation

Watson OpenScale monitors deployed models, compares predictions to actuals, and detects drift and quality issues over time. It is the platform service designed for ongoing model monitoring and evaluation.

Question 35Data and AI architecture

An architect wants to automate platform administration such as creating service instances and managing users. Which approach is supported in Cloud Pak for Data v4.7?

  1. Sending unauthenticated HTTP requests to service pods.
  2. Relying only on manual clicks in the web interface.
  3. Using the platform's REST APIs and command-line interface with authenticated credentials. Correct answer.
  4. Editing the internal database tables directly from a shell.

Explanation

The platform exposes REST APIs and a command-line interface for automation, and calls must carry authenticated credentials. Direct database edits bypass supported interfaces and can corrupt platform state.

Question 36Data and AI architecture

An architect must ensure a catalog asset with personal data is visible only to approved users while other assets remain broadly discoverable. Which design is correct?

  1. Restrict the entire catalog to a small group so no one else can discover any asset.
  2. Grant all users the steward role so they can decide for themselves.
  3. Apply asset-level access controls and membership rules so only approved users can view that asset, while leaving other assets open. Correct answer.
  4. Remove the asset from the catalog so no governance record exists.

Explanation

Cloud Pak for Data enforces access at the asset level, so architects can restrict sensitive assets to specific users or groups while keeping the rest of the catalog discoverable. This satisfies least-privilege without over-restricting unrelated assets.

Question 37Data and AI architecture

A regulated client requires that data remain in a specific country while a global team analyzes it through the platform. Which architecture pattern best meets this requirement?

  1. Allow analysts to download raw data to local laptops for analysis.
  2. Deploy platform instances in the required region and give the global team governed access to those regional instances. Correct answer.
  3. Copy all data to a central global cluster and delete the regional copies.
  4. Store only encrypted backups abroad and query them directly.

Explanation

Residency is preserved by keeping the platform and data in the required region and granting the distributed team governed access there. Copying data elsewhere or downloading it to endpoints breaks the residency constraint.

Question 38Data and AI architecture

An architect must design lineage reporting so a regulator can trace a reported metric back to its source columns. Which combination is required?

  1. Cataloged assets with captured lineage relationships plus governance metadata linking transformations to sources. Correct answer.
  2. Notebook comments written by the original developer.
  3. A glossary of business terms without any technical asset metadata.
  4. Only the final dashboard with no intermediate asset records.

Explanation

End-to-end lineage requires cataloged technical assets and captured lineage links across transformations, combined with governance metadata that ties business metrics to those assets. Without intermediate lineage records, tracing to source columns is not possible.

Question 39Data and AI architecture

A client wants generative AI assistants to answer questions over governed enterprise data while reducing the risk of exposing sensitive records. Which design decision best supports this?

  1. Load all enterprise data into an ungoverned vector store accessible to every user.
  2. Ground the assistant in governed, access-controlled assets so retrieval respects existing permissions and policies. Correct answer.
  3. Disable catalog governance so the assistant can reach more sources.
  4. Rely on prompt instructions telling the model to avoid sensitive data.

Explanation

Grounding assistants in governed assets means retrieval honors the platform's access controls and policies, so users see only data they are entitled to. Ungoverned stores and prompt-only safeguards do not enforce permissions.

Question 40Data and AI architecture

An architect is comparing a physical warehouse with a virtualized access layer for a reporting workload with strict freshness needs. Which consideration favors virtualization?

  1. Virtualization guarantees faster query performance than any warehouse.
  2. Queries read current source data at query time, avoiding staleness from load cycles. Correct answer.
  3. Virtualization removes the need for any access controls on sources.
  4. Virtualization permanently copies data into platform-managed storage.

Explanation

Virtualization queries sources at run time, so results reflect current data without waiting for load cycles, which suits strict freshness requirements. Performance and governance still depend on source capabilities and platform policies.

Question 41Platform installation, configuration, and upgrade

When planning a Cloud Pak for Data v4.7 installation on Red Hat OpenShift, which CLI tool is used to run the interactive installer that generates the custom resource manifests?

  1. cloudctl
  2. cpd-cli Correct answer.
  3. ibm-pak
  4. oc adm catalog mirror

Explanation

The cpd-cli tool is the supported command-line interface for Cloud Pak for Data v4.7 and provides commands such as cpd-cli manage to generate and apply the custom resources that drive the installation. It wraps the underlying OpenShift operations and records the components and versions being deployed.

Question 42Platform installation, configuration, and upgrade

Which OpenShift object must exist before the Cloud Pak for Data operator can be installed into a namespace?

  1. An ImageStream referencing the cpd operator catalog
  2. A pre-created Route with a custom TLS certificate
  3. A dedicated Project (namespace) with the required service accounts Correct answer.
  4. A PersistentVolumeClaim sized for the entire platform

Explanation

Cloud Pak for Data is installed into a dedicated project that the installer creates or reuses, and the operator needs its service accounts and RBAC in that namespace. Routes, PVCs, and ImageStreams are created later by the operator as part of the deployed services.

Question 43Platform installation, configuration, and upgrade

An architect needs to install Cloud Pak for Data in an air-gapped cluster. What must be done before running the installer?

  1. Disable the OpenShift image registry and rely on local node storage
  2. Enable the community operator catalog as the only source
  3. Mirror the required IBM operator catalogs and container images to a private registry reachable by the cluster Correct answer.
  4. Convert all persistent storage to hostPath volumes

Explanation

Air-gapped installations require the IBM operator catalogs and all referenced images to be mirrored into a registry the cluster can reach, with an ImageContentSourcePolicy or IDMS mapping the original references. The installer then references that private registry instead of the public one.

Question 44Platform installation, configuration, and upgrade

Which storage capability is required for a production Cloud Pak for Data installation that will host stateful services such as Db2 and Watson services?

  1. Node-local ephemeral storage only
  2. An NFS export mounted manually on every node
  3. Object storage used exclusively for all database volumes
  4. A dynamic provisioner that supports ReadWriteMany and ReadWriteOnce access modes with sufficient IOPS Correct answer.

Explanation

Stateful Cloud Pak services require persistent volumes backed by a supported storage class, typically needing both ReadWriteOnce and ReadWriteMany access modes and adequate performance. The installer validates storage classes before proceeding, so an unsupported provisioner blocks the installation.

Question 45Platform installation, configuration, and upgrade

What is the purpose of the Cloud Pak for Data control plane namespace versus the operand namespace in a v4.7 deployment?

  1. The control plane runs shared platform services and the operator, while operands run the individual service instances Correct answer.
  2. The control plane holds the registry and the operands hold the catalog sources
  3. The control plane stores only backups and the operands store only logs
  4. The control plane is used only during installation and is deleted afterward

Explanation

In the v4.7 architecture the control plane namespace hosts the operator and shared platform components such as the identity provider and zen services, while operand namespaces host the actual service instances. Separating them supports scaling and lifecycle management of individual services.

Question 46Platform installation, configuration, and upgrade

Which command-line interface does cpd-cli use under the hood to apply the generated custom resources to the cluster?

  1. docker
  2. helm
  3. oc Correct answer.
  4. kubectl only

Explanation

cpd-cli manage orchestrates installation by invoking oc commands against the target cluster, which is why a working oc login session and matching oc client are prerequisites. The installer relies on OpenShift-specific resources such as Routes and SecurityContextConstraints that plain kubectl does not fully manage.

Question 47Platform installation, configuration, and upgrade

Before running the Cloud Pak for Data installer, which cluster-level resource is commonly adjusted to allow the platform's pods to run with the required privileges?

  1. LimitRange
  2. PodDisruptionBudget
  3. SecurityContextConstraints Correct answer.
  4. ResourceQuota

Explanation

Cloud Pak for Data components may require specific SecurityContextConstraints to run with the user IDs and capabilities they expect. The installer checks and, where permitted, applies the necessary SCC bindings to the service accounts.

Question 48Platform installation, configuration, and upgrade

An architect is choosing an installation profile for Cloud Pak for Data v4.7. Which statement best describes the difference between the small and production profiles?

  1. The small profile omits the control plane entirely
  2. The small profile reduces resource requests for evaluation, while the production profile sizes components for sustained enterprise workloads Correct answer.
  3. The production profile installs only the operator without operands
  4. The small profile requires more nodes than the production profile

Explanation

Installation profiles adjust the resource requests and replica counts of platform components so that a small footprint can run on limited hardware for evaluation. The production profile allocates resources appropriate for sustained multi-user enterprise workloads.

Question 49Platform installation, configuration, and upgrade

Which component must be configured to provide identity and access management for a Cloud Pak for Data v4.7 installation?

  1. IBM Cloud Pak foundational services IAM Correct answer.
  2. A standalone LDAP server without integration
  3. An external Kerberos KDC only
  4. The OpenShift built-in OAuth only

Explanation

Cloud Pak for Data relies on IBM Cloud Pak foundational services for identity and access management, including user authentication and authorization. The foundational services are installed as part of the platform deployment and integrate with the chosen identity provider.

Question 50Platform installation, configuration, and upgrade

After installation, an architect needs to verify that all Cloud Pak for Data operators reached a healthy state. Which approach is most appropriate?

  1. Check the status conditions of the installed custom resources and the operator pods in the control plane namespace Correct answer.
  2. Inspect only the node kubelet logs
  3. Query the private registry catalog for image tags
  4. Review the OpenShift cluster version only

Explanation

Health verification centers on the custom resource status conditions and the state of operator and platform pods in the control plane namespace. These reflect whether reconciliation succeeded and whether dependent services are ready.