I85PREP
⌘ K
IBM watsonx Data Lakehouse Engineer Associate study materials
IBM watsonx Data Lakehouse Engineer Associate study resources

IBM watsonx Data Lakehouse Engineer Associate practice questions

One question at a time. Build your IBM watsonx Data Lakehouse Engineer Associate 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 1watsonx.data architecture and provisioning

In a watsonx.data deployment, which component stores table data and metadata for the Presto (PrestoDB) engine so that multiple engines can query the same tables?

  1. The Hive metastore plus an object store such as IBM Cloud Object Storage Correct answer.
  2. The Presto coordinator's local disk
  3. The analytics engine's in-memory cache
  4. The Db2 warehouse catalog only

Explanation

watsonx.data separates compute from storage: a metastore (Hive metastore) holds catalog and table definitions while actual data files live in object storage, so Presto, Presto C++, and Spark can all query the same tables. Local coordinator disk and engine memory are transient and not shared storage layers.

Question 2watsonx.data architecture and provisioning

Which statement best describes the role of the watsonx.data metastore in the lakehouse architecture?

  1. It maintains catalog definitions that map schemas and tables to their storage locations Correct answer.
  2. It schedules ingestion jobs and manages connector credentials
  3. It executes SQL queries and returns result sets to clients
  4. It compresses Parquet files before they are written to object storage

Explanation

The metastore is the metadata service: it records databases, schemas, tables, and the physical location of each table's files. Query execution is done by engines, and file compression is handled by the writer or table format, not the metastore.

Question 3watsonx.data architecture and provisioning

A team wants to provision a watsonx.data instance on IBM Cloud. Which deployment model should they expect for a fully managed service where IBM handles infrastructure patching?

  1. The IBM Cloud managed SaaS offering Correct answer.
  2. A self-managed cluster on their own Kubernetes nodes
  3. An on-premises OpenShift installation they administer
  4. A standalone Presto binary on a virtual machine

Explanation

watsonx.data is offered as a managed service on IBM Cloud where IBM operates the underlying infrastructure, and as a software deployment the customer runs on their own OpenShift environment. Self-managed Kubernetes or a bare Presto binary is not the managed SaaS model.

Question 4watsonx.data architecture and provisioning

Which engine in watsonx.data is best suited to low-latency, high-concurrency BI queries against data in object storage because it uses a vectorized C++ execution engine?

  1. Spark
  2. Presto (Java)
  3. Db2 Warehouse
  4. Presto C++ Correct answer.

Explanation

Presto C++ is the C++-based engine in watsonx.data built for fast, concurrent analytics on open data formats, complementing the Java Presto engine. Spark targets batch and ETL workloads, and Db2 is a relational warehouse rather than the vectorized query engine described.

Question 5watsonx.data architecture and provisioning

An administrator is adding a new object storage bucket as a data source in watsonx.data. What must be configured so tables in that bucket become queryable through the catalog?

  1. A Kafka topic subscribed to the bucket
  2. A new Presto worker node per bucket
  3. A storage registration with endpoint and credentials, plus a catalog that references it Correct answer.
  4. A Db2 backup image of the bucket

Explanation

To query data in object storage you register the storage (endpoint, bucket, and access credentials) and create a catalog that points at that storage and its metastore. Buckets do not require dedicated worker nodes, backups, or streaming topics to be queryable.

Question 6watsonx.data architecture and provisioning

Which watsonx.data component provides a unified SQL entry point so users can query across multiple catalogs without knowing which engine or storage holds each table?

  1. The Hive metastore
  2. The ingestion pipeline scheduler
  3. The object storage bucket policy
  4. The query workspace layer that routes through configured engines and catalogs Correct answer.

Explanation

watsonx.data exposes a query workspace where users run SQL against catalogs, and the service routes the statement to an appropriate engine with access to that catalog. The metastore stores metadata, bucket policies govern access, and ingestion scheduling is a separate function.

Question 7watsonx.data architecture and provisioning

A data engineer needs to create a schema in watsonx.data that points to an existing Hive metastore used by another team. Which action accomplishes this?

  1. Copy every table into a new bucket before querying
  2. Attach the metastore as a Presto worker
  3. Register the existing metastore as a catalog and associate it with the shared storage Correct answer.
  4. Convert the tables to Db2 format first

Explanation

watsonx.data lets you register an external Hive metastore as a catalog together with its backing storage, so existing table definitions are immediately visible without copying data. Copying, worker attachment, or conversion would duplicate or alter data unnecessarily.

Question 8watsonx.data architecture and provisioning

What is the primary benefit of watsonx.data's separation of compute and storage when a workload spikes?

  1. Metadata is rewritten on every query
  2. Each query is forced onto a single fixed node
  3. Storage capacity automatically shrinks to match demand
  4. Engines can be scaled independently without moving or duplicating the data Correct answer.

Explanation

Because data resides in shared object storage and compute runs in separate engines, you can add or resize engine capacity during demand spikes without relocating data. Storage does not shrink automatically, and queries are distributed rather than pinned to one node.

Question 9watsonx.data architecture and provisioning

Which data format is most commonly recommended for tables stored in watsonx.data object storage to enable efficient columnar analytics?

  1. XML
  2. Parquet Correct answer.
  3. Plain text logs
  4. CSV

Explanation

Parquet is a columnar, compressed format that lets engines read only needed columns, which suits analytic workloads on object storage. CSV, plain text, and XML are row-oriented or verbose and less efficient for large-scale analytics.

Question 10watsonx.data architecture and provisioning

An administrator must grant a new analyst the ability to run queries in watsonx.data without allowing them to register storage or alter catalogs. Which approach fits the platform's access model?

  1. Disable authentication for the instance
  2. Share the administrator's credentials with the analyst
  3. Assign the analyst an administrative role
  4. Assign a query-oriented role scoped to the relevant catalogs Correct answer.

Explanation

watsonx.data uses role-based access so users can be given query privileges on specific catalogs without administrative rights over storage or catalog configuration. Administrative roles, shared credentials, and disabling authentication all violate least-privilege practice.

Question 11watsonx.data architecture and provisioning

Which statement accurately distinguishes the Java Presto engine from Presto C++ in a watsonx.data deployment?

  1. Presto C++ requires a Db2 backend
  2. Java Presto cannot read Parquet files
  3. Java Presto only supports Hive metastore version 1
  4. Presto C++ is a native compiled engine while Java Presto runs on the JVM Correct answer.

Explanation

Both engines query open-format data, but Presto C++ is a natively compiled engine whereas the original Presto runs on the Java Virtual Machine. Both can read Parquet, and neither mandates a Db2 backend or a specific legacy metastore version.

Question 12watsonx.data architecture and provisioning

A team is provisioning watsonx.data and must decide where query result data and temporary spill files go. Which consideration is most relevant?

  1. Temporary space is unnecessary because queries never spill
  2. Spill files must be stored in the metastore database
  3. All temporary data should be written to the shared object bucket
  4. Spill and temporary space should be provisioned on fast local storage attached to the engine Correct answer.

Explanation

Engines use local disk for temporary and spill files during large operations, so fast local storage sized for peak workloads matters for performance. Temporary data is not written to the shared object bucket or the metastore, and queries can spill when memory is exceeded.

Question 13watsonx.data architecture and provisioning

Which capability allows watsonx.data to query data that remains in an external Db2 database without first copying it into object storage?

  1. Attaching the Db2 instance as a Presto worker
  2. Exporting Db2 to CSV and loading it into Hive
  3. Federation through a configured database connection and catalog Correct answer.
  4. Converting Db2 tables to Parquet nightly

Explanation

watsonx.data can register an external relational database as a catalog and federate queries to it, so data stays in place while being joined with lakehouse tables. Copying, converting, or attaching the database as a worker is not how federation works.

Question 14watsonx.data architecture and provisioning

What does registering a bucket in watsonx.data typically require beyond the bucket name?

  1. A Presto worker ID
  2. A Hive table definition for every file
  3. An endpoint URL and access credentials valid for that bucket Correct answer.
  4. A Db2 license key

Explanation

Storage registration needs the endpoint for the object store plus credentials (such as an API key or access key) that authorize access to the bucket. Worker IDs, per-file table definitions, and Db2 licenses are not part of bucket registration.

Question 15watsonx.data architecture and provisioning

A user reports that a table created through watsonx.data is visible in one catalog but not another. What is the most likely explanation?

  1. The table was automatically moved to Db2
  2. The table's metadata was written to the metastore associated with the first catalog only Correct answer.
  3. Object storage buckets can only hold one table
  4. Presto C++ deletes tables created by Java Presto

Explanation

Each catalog is backed by a specific metastore, so a table defined in one metastore is only visible through catalogs referencing that metastore. Buckets can hold many tables, engines do not delete each other's tables, and tables are not auto-migrated to Db2.

Question 16watsonx.data architecture and provisioning

Which statement about the watsonx.data query workspace is accurate?

  1. It provides a browser-based SQL editor where users select a catalog and engine Correct answer.
  2. It only supports Spark SQL
  3. It bypasses catalog permissions entirely
  4. It requires a separate client install for each engine

Explanation

The query workspace is a web-based SQL editor in watsonx.data where users choose the engine and catalog for a statement. It does not require per-engine client installs, supports SQL across engines, and still honors catalog permissions.

Question 17watsonx.data architecture and provisioning

An architect wants to isolate a heavy ETL workload from interactive BI users. Which watsonx.data design choice best achieves this?

  1. Run ETL on a separate engine instance from the BI engine, both reading the same storage Correct answer.
  2. Move BI data into the metastore
  3. Put both workloads on one engine and lower the BI query timeout
  4. Store ETL output only on the coordinator disk

Explanation

Because compute is decoupled from storage, you can provision distinct engines for ETL and BI so resource contention is avoided while both read the same object storage. Timeouts do not isolate resources, the metastore cannot hold analytic data, and coordinator disk is not durable shared storage.

Question 18watsonx.data architecture and provisioning

Which open table format support in watsonx.data enables ACID transactions and time travel on object storage tables?

  1. Plain CSV directories
  2. Gzip-compressed text
  3. Delta Lake and Iceberg Correct answer.
  4. Db2 export files

Explanation

Open table formats such as Delta Lake and Apache Iceberg add transactional metadata, schema evolution, and time travel on top of object storage files. CSV, gzip text, and Db2 exports lack these transactional table semantics.

Question 19watsonx.data architecture and provisioning

During provisioning, an administrator must ensure the watsonx.data instance can reach an on-premises object store. What is the key prerequisite?

  1. Installing Presto on the object store host
  2. Disabling TLS on the endpoint
  3. Converting the store to a relational database
  4. Network connectivity such as a VPN or private link between the instance and the store Correct answer.

Explanation

Registering an on-premises or private object store requires network reachability, typically through a VPN, private link, or similar connectivity, plus valid credentials. Installing Presto on the store, converting it to a database, or disabling TLS is neither required nor advisable.

Question 20watsonx.data architecture and provisioning

Which watsonx.data component would an administrator consult to confirm which storage location backs a specific table?

  1. The object storage bucket policy
  2. The ingestion job history
  3. The catalog entry in the metastore Correct answer.
  4. The Presto coordinator log only

Explanation

The catalog entry in the metastore records the table's schema and its physical storage location, so it is the authoritative place to confirm the backing path. Logs, ingestion history, and bucket policies do not define table-to-storage mappings.

Question 21watsonx.data architecture and provisioning

What advantage does using watsonx.data's managed SaaS deployment provide over a self-managed installation?

  1. Customers gain root access to the underlying hosts
  2. Storage costs are eliminated
  3. Engines no longer require catalogs
  4. IBM handles infrastructure provisioning and maintenance of the service Correct answer.

Explanation

In the managed SaaS model IBM provisions and maintains the underlying infrastructure, reducing operational burden for customers. Customers do not get root access to hosts, storage still incurs cost, and catalogs remain necessary to query data.

Question 22watsonx.data architecture and provisioning

A table's files were written by Spark into a bucket, and an analyst wants to query them with Presto in watsonx.data. What must exist for Presto to see the table?

  1. A Db2 synonym for the table
  2. A new Spark cluster dedicated to the analyst
  3. A copy of the files in the Presto coordinator directory
  4. A catalog whose metastore contains the table definition pointing at those files Correct answer.

Explanation

Presto locates tables through a catalog backed by a metastore that records the table definition and file location. Copying files locally, creating a Db2 synonym, or spinning up another Spark cluster does not make the table visible to Presto.

Question 23watsonx.data architecture and provisioning

Which factor most directly determines the number of Presto worker nodes needed for a watsonx.data engine handling concurrent dashboards?

  1. The size of the metastore database file
  2. Expected concurrent query load and data volume scanned per query Correct answer.
  3. The count of catalogs defined in the instance
  4. The number of buckets registered in storage

Explanation

Worker sizing is driven by concurrency and the volume of data each query must scan, since those determine CPU and memory demand. Bucket count, metastore size, and catalog count are metadata concerns that do not directly set compute capacity.

Question 24watsonx.data architecture and provisioning

Which statement about watsonx.data catalogs is correct?

  1. A catalog can only reference a single table
  2. A catalog replaces the need for object storage
  3. A catalog associates a metastore with a storage location so tables can be queried Correct answer.
  4. A catalog is a physical copy of all table data

Explanation

A catalog binds a metastore to the storage it manages, giving engines a namespace in which to resolve tables. It is not a copy of the data, is not limited to one table, and does not remove the need for underlying storage.

Question 25watsonx.data architecture and provisioning

An organization must keep lakehouse data in its own data center while using watsonx.data. Which deployment option matches this requirement?

  1. Use the IBM Cloud managed SaaS instance
  2. Use only the query workspace without an instance
  3. Deploy watsonx.data software on their own OpenShift cluster Correct answer.
  4. Rely on a shared public bucket

Explanation

For data-residency requirements, watsonx.data can be deployed as software on the customer's own OpenShift environment, keeping data on-premises. The managed SaaS runs in IBM Cloud, a shared public bucket does not provide the platform, and the query workspace requires an instance.

Question 26watsonx.data architecture and provisioning

In a watsonx.data deployment, which layer is responsible for storing the actual table data files while the query engines remain stateless between runs?

  1. The metastore catalog service
  2. The Presto coordinator's local disk
  3. The object storage layer, such as an S3-compatible bucket Correct answer.
  4. The browser-based query workspace

Explanation

watsonx.data separates storage from compute, so persistent table data lives in object storage buckets that engines read on demand. Engines hold only transient state such as in-memory buffers and spill files. The metastore stores metadata, not the data files themselves.

Question 27watsonx.data architecture and provisioning

Which watsonx.data component stores metadata such as database, schema, and table definitions so engines can resolve a table name to its underlying files?

  1. The metastore catalog Correct answer.
  2. The object storage bucket
  3. The Presto worker pool
  4. The ingestion pipeline

Explanation

The metastore catalog holds the metadata that maps names to physical locations, and engines consult it when planning queries. Buckets hold data, workers execute query fragments, and ingestion pipelines move data. Without the catalog, engines could not resolve table names.

Question 28watsonx.data architecture and provisioning

An administrator is creating a new watsonx.data instance and must choose an engine for ad hoc SQL over Iceberg tables. Which engine type is designed specifically for interactive, low-latency SQL analytics in watsonx.data?

  1. Spark
  2. A Kafka broker
  3. Presto Correct answer.
  4. Db2 Warehouse

Explanation

Presto is the SQL query engine in watsonx.data built for interactive analytics across federated sources. Spark is used for batch transformation and ETL rather than low-latency interactive SQL. Db2 Warehouse is a relational database, and Kafka is a streaming platform.

Question 29watsonx.data architecture and provisioning

When provisioning watsonx.data, an administrator wants the service to manage infrastructure patching and scaling automatically. Which deployment model provides that?

  1. A local developer sandbox on a laptop
  2. A container image run on an on-premises VM
  3. A self-managed installation on customer-managed Kubernetes
  4. The fully managed SaaS deployment on IBM Cloud Correct answer.

Explanation

The managed SaaS offering has IBM operate the underlying infrastructure, including patching and scaling. Self-managed, sandbox, and on-premises options shift operational responsibility to the customer. Choosing SaaS reduces administrative overhead for the team.

Question 30watsonx.data architecture and provisioning

Which statement best describes the role of a catalog in watsonx.data?

  1. It encrypts data at rest in object storage buckets
  2. It defines a namespace of databases and tables that engines can query through a single interface Correct answer.
  3. It replaces object storage by holding table rows in memory
  4. It schedules periodic backups of the metastore

Explanation

A catalog is a named namespace registered in watsonx.data that exposes databases and tables to the query engines. Multiple catalogs can be attached so users query across sources consistently. Catalogs organize metadata references; they do not encrypt data, replace storage, or run backups.

Question 31watsonx.data architecture and provisioning

A data engineer wants watsonx.data to read Parquet files already sitting in an object storage bucket without moving them. Which action is required first?

  1. Load the files into the metastore as rows
  2. Convert the files to CSV
  3. Create a Presto worker for each file
  4. Register the bucket as a storage in watsonx.data Correct answer.

Explanation

Before tables over existing files can be queried, the storage location must be registered so watsonx.data knows how to reach it. Registration captures connection details such as endpoint and credentials. Converting formats or loading rows is unnecessary because the files are queried in place.

Question 32watsonx.data architecture and provisioning

Which watsonx.data capability lets a single SQL statement join a table in the lakehouse with a table in an external relational database?

  1. Schema evolution on Parquet files
  2. Federation through a registered database connection Correct answer.
  3. Time travel on Iceberg tables
  4. Bucket lifecycle policies

Explanation

Federation lets the engine push queries to external databases and combine results with lakehouse tables in one statement. Time travel and schema evolution concern table history and structure, not cross-source joins. Lifecycle policies manage object retention in storage.

Question 33watsonx.data architecture and provisioning

An administrator must confirm that a newly registered storage in watsonx.data is reachable and credentials are valid. Which step verifies this during setup?

  1. Disable encryption on the bucket
  2. Restart all Presto workers
  3. Delete and recreate the metastore
  4. Test the storage connection after entering endpoint and access credentials Correct answer.

Explanation

Registration captures endpoint, region, and credential details, and a connection test confirms watsonx.data can reach the bucket with the supplied credentials. This catches typos and permission problems before tables are created. Deleting the metastore, restarting workers, or disabling encryption do not validate connectivity.

Question 34watsonx.data architecture and provisioning

A team needs to keep lakehouse data inside its own data center while still using watsonx.data software. Which deployment approach fits this requirement?

  1. A public cloud bucket shared with IBM
  2. A self-managed watsonx.data installation on infrastructure the organization controls Correct answer.
  3. The managed SaaS deployment hosted by IBM
  4. A read-only replica in a third-party region

Explanation

A self-managed installation runs watsonx.data on infrastructure the organization operates, which keeps data resident in its own data center. Managed SaaS places operational control with IBM, and sharing buckets or replicas elsewhere moves data outside the data center. Data residency requirements therefore point to self-managed deployment.

Question 35watsonx.data architecture and provisioning

An engineer creates a schema inside a watsonx.data catalog and points it at a new bucket prefix. Which object ultimately records where the tables in that schema live?

  1. The Presto coordinator's configuration file
  2. The query workspace history panel
  3. The metastore catalog entry for the schema and its tables Correct answer.
  4. The object storage access log

Explanation

The metastore catalog records schema and table definitions including their storage locations, so engines can resolve names to files. Coordinator configuration governs engine behavior, not table locations. Workspace history and access logs are operational records, not metadata sources.

Question 36watsonx.data architecture and provisioning

A workload runs large nightly transformations and also serves daytime interactive dashboards. Which watsonx.data design best isolates these workloads from each other?

  1. Disable the metastore during batch runs
  2. Move all dashboards to run against the transformation engine
  3. Use one engine with a larger coordinator for both workloads
  4. Provision separate engines for batch transformation and interactive SQL Correct answer.

Explanation

Because compute is decoupled from storage, multiple engines can read the same data while keeping resources isolated. Separate engines prevent a heavy batch job from starving interactive queries. A single shared engine or routing dashboards to the batch engine reintroduces contention, and disabling the metastore breaks metadata resolution.

Question 37watsonx.data architecture and provisioning

An analyst queries a table and gets a permission error even though the table exists in the catalog. Which watsonx.data layer most likely needs adjustment?

  1. The Parquet compression codec
  2. The bucket's object versioning setting
  3. The Presto worker count
  4. Access control policies governing the catalog and its data Correct answer.

Explanation

Permission errors point to access control governing catalogs, schemas, and data. Compression codecs, worker counts, and versioning affect performance or storage behavior, not authorization. Adjusting the relevant access policy resolves the error.

Question 38watsonx.data architecture and provisioning

A team wants ACID transactions and time travel on tables stored in object storage within watsonx.data. Which table format should they choose?

  1. Avro
  2. Plain CSV
  3. JSON
  4. Iceberg Correct answer.

Explanation

Iceberg maintains snapshots and metadata that enable ACID semantics and time travel on object storage. CSV, Avro, and JSON are file formats without table-level transaction or snapshot management. Choosing Iceberg gives the required transactional behavior.

Question 39watsonx.data architecture and provisioning

An administrator must give a data scientist the ability to run queries without permitting creation of new catalogs or storage registrations. Which approach follows least privilege in watsonx.data?

  1. Grant full administrator privileges to simplify setup
  2. Assign a role scoped to querying existing catalogs and data Correct answer.
  3. Share the administrator credentials with the scientist
  4. Register the scientist as a storage owner

Explanation

Least privilege means granting only the permissions needed, so a query-scoped role fits the task. Administrator rights, shared credentials, and storage ownership exceed the requirement and widen risk. Role-based scoping keeps catalog and storage changes restricted.

Question 40watsonx.data architecture and provisioning

A Presto engine in watsonx.data is running out of local disk during large sorts. Which configuration should the administrator review?

  1. The bucket's public access setting
  2. The metastore schema naming convention
  3. The spill directory and its available capacity Correct answer.
  4. The number of catalogs registered

Explanation

Large sorts and aggregations spill intermediate data to disk when memory is exhausted, so the spill location and capacity matter. Catalog count, bucket access settings, and schema naming do not affect spill capacity. Ensuring adequate spill storage prevents query failures.

Question 41watsonx.data architecture and provisioning

An organization must query data that stays in an on-premises object store while using watsonx.data. What is the key provisioning consideration?

  1. Disabling the metastore to avoid conflicts
  2. Using only the query workspace without engines
  3. Converting all data to a relational database first
  4. Network connectivity and credentials that let watsonx.data reach the on-premises endpoint Correct answer.

Explanation

Reaching an on-premises object store requires network paths and valid credentials so watsonx.data can register and read the location. Converting data or disabling the metastore is unnecessary and disruptive, and queries still require an engine. Connectivity and credentials are the gating factors.

Question 42watsonx.data architecture and provisioning

A user reports that a table created in one catalog cannot be found when querying from another catalog. What is the most likely explanation?

  1. The Presto coordinator needs a larger heap
  2. The table's Parquet files were compressed incorrectly
  3. The bucket has versioning enabled
  4. The table exists only in the namespace of the catalog where it was created Correct answer.

Explanation

Catalogs define separate namespaces, so a table created in one catalog is not automatically visible in another. Cross-catalog access requires referencing the correct catalog or registering the location elsewhere. Compression, heap size, and versioning do not determine namespace visibility.

Question 43watsonx.data architecture and provisioning

An architect is sizing a watsonx.data engine for unpredictable concurrency. Which factor most directly drives the number of worker nodes required?

  1. Peak concurrent query count and per-query memory needs Correct answer.
  2. The storage class of the bucket
  3. The number of buckets registered in the instance
  4. The count of catalogs defined

Explanation

Worker capacity must cover peak concurrent queries and the memory each query consumes. Bucket count, catalog count, and storage class affect metadata scope or cost rather than compute concurrency. Sizing on peak concurrency and memory avoids both under-provisioning and waste.

Question 44watsonx.data architecture and provisioning

A team wants a single SQL endpoint that can query tables across several catalogs without users knowing which engine or catalog holds each table. Which watsonx.data feature supports this?

  1. A nightly copy of all tables into one schema
  2. A dedicated bucket per user
  3. A query engine that federates across registered catalogs Correct answer.
  4. Manual export of each table to CSV

Explanation

Federated querying through one engine lets users write SQL against multiple catalogs without tracking physical locations. Per-user buckets, CSV exports, and nightly copies add complexity and duplication. Federation provides the unified access the team wants.

Question 45watsonx.data architecture and provisioning

An administrator is deciding between Presto and Spark for a watsonx.data workload that involves complex multi-step transformations written in Python. Which choice fits best?

  1. Spark, because it supports programmatic transformation logic alongside SQL Correct answer.
  2. Spark, because it cannot run SQL at all
  3. Presto, because it is the only engine that can read Iceberg
  4. Presto, because it stores data permanently on workers

Explanation

Spark supports programmatic data processing in languages such as Python in addition to SQL, making it suited to multi-step transformations. Presto is optimized for interactive SQL and does not persist data on workers. Both engines can read Iceberg tables.

Question 46watsonx.data architecture and provisioning

During provisioning, an administrator must decide where query result sets and temporary files are stored. Which consideration applies?

  1. They are stored permanently in the same bucket as source tables
  2. These are transient artifacts placed on compute-side storage, not in the persistent lakehouse buckets Correct answer.
  3. They are kept only in the user's browser
  4. They must be written into the metastore catalog

Explanation

Result sets and spill files are transient and belong to compute-side storage, separate from persistent lakehouse data. Writing them into the metastore or source buckets mixes temporary artifacts with durable data. Browser storage cannot hold engine working files.

Question 47watsonx.data architecture and provisioning

A company needs to add a new analytics engine without copying data or causing downtime for existing queries. Which watsonx.data characteristic makes this possible?

  1. Mandatory data duplication per engine
  2. A single shared coordinator for all engines
  3. Decoupled storage so a new engine can attach to the same catalogs and buckets Correct answer.
  4. Exclusive bucket ownership per engine

Explanation

Because data lives in shared object storage and metadata in catalogs, a new engine can be provisioned to read the same tables without moving data. Shared coordinators, duplication, and exclusive ownership would undermine this flexibility. Decoupling is what enables elastic engine addition.

Question 48watsonx.data architecture and provisioning

An engineer registers an external Hive metastore as a catalog in watsonx.data. Which outcome should be expected?

  1. The external metastore is replaced by the watsonx.data metastore
  2. The metastore's tables are automatically copied into watsonx.data buckets
  3. All Presto workers are restarted to pick up the tables
  4. Tables defined in that metastore become queryable through watsonx.data without relocating their files Correct answer.

Explanation

Registering an external metastore exposes its table definitions to watsonx.data so the engines can query the existing files in place. It does not copy data, replace the external metastore, or require worker restarts. The catalog acts as a bridge to existing metadata.

Question 49watsonx.data architecture and provisioning

A security review asks how to limit which users can register new storage in watsonx.data while still letting them query existing tables. Which design meets this?

  1. Remove all roles and rely on network controls
  2. Grant every user storage registration rights for convenience
  3. Give all users administrator credentials shared by the team
  4. Separate roles so storage registration is restricted to administrators and querying is granted broadly Correct answer.

Explanation

Separating administrative capabilities such as storage registration from query access enforces least privilege. Granting registration broadly, removing roles, or sharing administrator credentials removes necessary controls. Role separation keeps storage changes restricted while preserving query access.

Question 50watsonx.data architecture and provisioning

A team observes that queries against a large Iceberg table are slow while a small table in the same catalog is fast. Which analysis is most reasonable before changing provisioning?

  1. Conclude that object storage cannot serve large tables
  2. Assume the catalog is misconfigured and recreate it
  3. Check table statistics and file layout, since engine sizing alone may not explain the difference Correct answer.
  4. Add more catalogs to distribute the load

Explanation

Performance differences between tables in the same catalog often stem from data layout, file sizes, and statistics rather than engine capacity. Recreating the catalog, blaming object storage, or adding catalogs does not address table-level factors. Reviewing statistics and layout is the sound first step.