I85PREP
⌘ K
Informatica Cloud DI Foundation study materials
Informatica Cloud DI Foundation study resources

Informatica Cloud DI Foundation practice questions

One question at a time. Build your Informatica Cloud DI Foundation 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 1Cloud Data Integration

In Informatica Cloud Data Integration, which component is the reusable, saved set of source-to-target transformation logic that a Mapping task executes?

  1. A Runtime Environment
  2. A Mapping Correct answer.
  3. A Connection
  4. A Schedule

Explanation

A Mapping is the design-time object that holds sources, transformations, and targets and is what a Mapping task runs. Connections only store access credentials and endpoints, while Runtime Environments and Schedules control where and when execution happens.

Question 2Cloud Data Integration

Which Secure Agent component is responsible for actually executing Mapping tasks in an organization's own network?

  1. The Administrator role
  2. The Secure Agent service installed on a local machine Correct answer.
  3. The Metadata Manager
  4. The Informatica Cloud Hosted Agent

Explanation

The Secure Agent is lightweight software installed on a machine in your environment; its service executes tasks and moves data locally. The Hosted Agent runs in Informatica's cloud, and Metadata Manager and roles are governance and security concepts rather than execution engines.

Question 3Cloud Data Integration

A developer needs to read only rows where ORDER_AMOUNT is greater than 500 before any other transformation runs. Which configuration achieves this most efficiently?

  1. Add a Filter transformation with the condition ORDER_AMOUNT > 500 directly after the Source Correct answer.
  2. Add an Expression transformation and set other rows to null
  3. Add an Aggregator transformation grouped by ORDER_AMOUNT
  4. Add a Router transformation with no groups defined

Explanation

A Filter transformation placed immediately after the Source discards non-matching rows early, reducing downstream processing. Setting values to null keeps the rows in the flow, and an Aggregator changes granularity rather than filtering.

Question 4Cloud Data Integration

Which task type should be used to run a Mapping on a recurring schedule without writing custom code?

  1. A Mapping task Correct answer.
  2. A Synchronization task
  3. A Mass Ingestion task
  4. A Data Assessment task

Explanation

A Mapping task runs a saved Mapping and can be scheduled and monitored in the operational console. Synchronization tasks are for simple point-to-point replication, and Data Assessment tasks profile data rather than execute mapping logic.

Question 5Cloud Data Integration

Which join type returns every row from the master source plus matching rows from the detail source, filling unmatched detail fields with nulls?

  1. Left outer join with the master as the left source Correct answer.
  2. Right outer join with the master as the right source
  3. Full outer join
  4. Inner join

Explanation

A left outer join keeps all rows from the left (master) source and appends matching detail values, leaving nulls where no match exists. An inner join drops unmatched master rows, and a full outer join would also retain unmatched detail rows.

Question 6Cloud Data Integration

In a Mapping, what does a Source transformation represent?

  1. A reusable parameter file
  2. The target table's primary key
  3. A physical database index
  4. The definition of the input data object and its fields Correct answer.

Explanation

The Source transformation declares the input object, its connection, and the fields that flow downstream. It does not store parameters or enforce target constraints, and indexes are database objects rather than mapping components.

Question 7Cloud Data Integration

Which statement best describes the role of a parameter file in a Mapping task?

  1. It stores the mapping's transformation logic
  2. It defines the target table's schema permanently
  3. It replaces the need for a Secure Agent
  4. It supplies values at run time so the same Mapping can serve multiple environments Correct answer.

Explanation

Parameter files externalize values such as connection names or file paths so one Mapping can be reused across environments without edits. They do not contain transformation logic, provide compute, or alter target schemas.

Question 8Cloud Data Integration

A pipeline must combine two differently structured sources into one target by matching field names automatically. Which transformation is designed for this?

  1. A Sequence transformation
  2. An Unstructured Data transformation
  3. A Lookup transformation
  4. A Union transformation Correct answer.

Explanation

A Union transformation merges multiple input pipelines into a single output, aligning fields by name or position. Lookups enrich rows, Sequence generates numeric keys, and Unstructured Data handles documents rather than tabular merging.

Question 9Cloud Data Integration

A developer wants to reject rows that fail a data quality rule and write them to a separate error target while valid rows continue. Which approach fits best?

  1. Use a Sequence transformation to number the rows
  2. Use a Router transformation with a valid group and a default group Correct answer.
  3. Use an Aggregator to count the bad rows
  4. Use a Joiner with a full outer join

Explanation

A Router transformation routes rows into different output groups based on conditions, so valid and invalid rows can go to separate targets. Aggregators summarize, Sequences number rows, and Joiners combine datasets rather than split them by rule.

Question 10Cloud Data Integration

Which transformation would you use to add a monotonically increasing surrogate key to each output row?

  1. Expression
  2. Sequence Correct answer.
  3. Rank
  4. Sorter

Explanation

The Sequence transformation generates sequential numeric values that can be assigned to a field as a surrogate key. Expression computes values from existing fields, Rank orders rows, and Sorter sorts them.

Question 11Cloud Data Integration

What is the primary purpose of a Lookup transformation in a Mapping?

  1. To split one row into multiple rows
  2. To enforce a primary key on the target
  3. To sort the incoming data
  4. To retrieve related values from a reference object and add them to the row Correct answer.

Explanation

A Lookup fetches values from a lookup source, such as a table or file, and returns them into the pipeline for enrichment. Sorting, row explosion, and target key enforcement are handled by other transformations or the target itself.

Question 12Cloud Data Integration

Which lookup configuration returns a value only when exactly one match is found and fails the row when multiple matches exist?

  1. Lookup with a dynamic cache
  2. Lookup with the 'Report Error' behavior on multiple matches Correct answer.
  3. Lookup with multiple matches allowed
  4. Unconnected lookup with a static cache

Explanation

Configuring the lookup to report an error when multiple matches occur enforces the one-match expectation and surfaces data issues. Allowing multiple matches returns an arbitrary row, and cache settings affect performance rather than match cardinality rules.

Question 13Cloud Data Integration

A Mapping reads a flat file whose delimiter and text qualifier vary by region. Which feature lets these be changed at run time without editing the Mapping?

  1. Enabling row-level logging
  2. Adding a Sorter transformation
  3. Using parameters for file format properties Correct answer.
  4. Hard-coding the delimiter in the source object

Explanation

File format properties such as delimiter and text qualifier can be parameterized so the same Mapping handles different regional files. Hard-coding fixes one format, and sorting or logging do not affect parsing.

Question 14Cloud Data Integration

Which statement about the Expression transformation is accurate?

  1. It replaces the need for a Filter transformation
  2. It can create output fields calculated from input fields and variables Correct answer.
  3. It can only rename fields
  4. It permanently modifies the source table

Explanation

An Expression transformation computes new output ports from input ports, local variables, and built-in functions. It cannot alter source tables, does more than renaming, and cannot drop rows the way a Filter does.

Question 15Cloud Data Integration

A developer needs the top three highest-value orders per customer. Which transformation combination is most appropriate?

  1. Sorter on amount followed by a Rank transformation grouped by customer Correct answer.
  2. Router with a single group
  3. Joiner with an inner join on customer
  4. Aggregator with SUM on amount

Explanation

Sorting by amount and then ranking within each customer group yields the top N rows per group. An Aggregator would collapse rows into totals, while Joiners and Routers do not produce ranked top-N subsets.

Question 16Cloud Data Integration

Which caching strategy is generally best when a lookup reference table is small and does not change during the run?

  1. Static cache Correct answer.
  2. Uncached with a database join each row
  3. No cache
  4. Dynamic cache

Explanation

A static cache loads the reference data once and reuses it for every row, which is efficient for small, unchanging tables. Dynamic caches are for changing reference data, and no cache forces a lookup per row.

Question 17Cloud Data Integration

What is the effect of setting the tracing level to Verbose in a Mapping task?

  1. It changes the target table's structure
  2. It encrypts the data in transit
  3. It disables the Secure Agent
  4. It increases the detail written to session logs Correct answer.

Explanation

Tracing levels control how much diagnostic information the task writes to its logs, with Verbose providing the most detail. They do not alter target structure, affect agent availability, or govern encryption.

Question 18Cloud Data Integration

A Mapping task fails with a 'connection refused' error when accessing an on-premises database. Which area should be investigated first?

  1. The target field data types
  2. The mapping's parameter file naming convention
  3. The Secure Agent host's network access and database listener Correct answer.
  4. The schedule's time zone

Explanation

Connection refused errors typically indicate the agent cannot reach the database endpoint, so network routing, firewall rules, and the database listener are the first checks. Data types, parameter naming, and schedule time zones do not cause connection refusals.

Question 19Cloud Data Integration

Which feature allows a Mapping to process source rows in parallel by partitioning data across threads?

  1. Using a single-threaded session
  2. Disabling the lookup cache
  3. Setting the tracing level to Verbose
  4. Partitioning configured on the pipeline Correct answer.

Explanation

Partitioning splits the data flow so multiple threads can process subsets concurrently, improving throughput on large volumes. Tracing, single-threaded sessions, and cache settings do not create parallelism.

Question 20Cloud Data Integration

A developer must load only new and changed rows from a source table into a target on each run. Which approach is most appropriate?

  1. Loading only the first 100 rows
  2. Incremental load using a watermark column such as last-modified timestamp Correct answer.
  3. Full reload of the target every run
  4. Random sampling of source rows

Explanation

Incremental loading uses a watermark such as a last-modified timestamp or increasing key to select only new or changed rows since the prior run. Full reloads reprocess everything, and sampling or row limits do not capture changes reliably.

Question 21Cloud Data Integration

Which statement about the Aggregator transformation's group by ports is correct?

  1. Group by ports must be numeric
  2. Group by ports are optional and have no effect
  3. Group by ports define the granularity of the aggregate output Correct answer.
  4. Group by ports replace the need for a Sorter

Explanation

Group by ports determine the level at which aggregate functions such as SUM or COUNT are computed. They can be any data type, do not substitute for sorting, and materially change the output when specified.

Question 22Cloud Data Integration

A Mapping joins a large fact source with a small dimension source. To reduce memory pressure on the Secure Agent, which Joiner configuration is preferable?

  1. Designate the small dimension as the master source for a nested-loop style join with caching Correct answer.
  2. Sort the fact source and use a sorted join
  3. Disable all caching and perform a block nested loop
  4. Use a full outer join

Explanation

When one side is small, making it the master allows the agent to cache it and stream the large detail side, reducing memory pressure. Full outer joins retain unmatched rows from both sides, and disabling caching increases work per row.

Question 23Cloud Data Integration

A developer needs to detect and capture deleted source rows that no longer exist in the source table during incremental loads. Which technique is appropriate?

  1. Use a Router with a single group
  2. Add a Sequence transformation
  3. Compare source keys against target keys and flag missing ones as deletes Correct answer.
  4. Increase the tracing level

Explanation

Soft-delete detection compares the current source key set with the target key set and marks target rows absent from the source as deleted. Tracing, sequencing, and single-group routing do not identify missing keys.

Question 24Cloud Data Integration

Which statement best explains why a dynamic lookup cache is required when the same run inserts new reference rows that later rows must match?

  1. The cache stores only the first matched row
  2. The cache is disabled to force database lookups
  3. The cache is refreshed from the lookup source after each insert so subsequent rows can match new values Correct answer.
  4. The cache converts all lookup values to strings

Explanation

A dynamic lookup cache is updated with rows inserted or updated during the run, so later rows can match the new reference values without a separate reload. Static caches would miss those new values until the next run.

Question 25Cloud Data Integration

A Mapping task intermittently fails with out-of-memory errors during a large aggregation. Which remediation is most likely to help?

  1. Set the tracing level to Verbose
  2. Increase the aggregation cache size or add partitioning to reduce per-thread data Correct answer.
  3. Remove the group by ports
  4. Convert the Aggregator to a Filter

Explanation

Aggregations hold group state in memory, so enlarging the aggregation cache or partitioning the data reduces per-thread memory pressure. Verbose tracing adds overhead, and removing grouping or replacing with a Filter changes the required logic.

Question 26Cloud Data Integration

In a Mapping, which object defines where the integrated data is written after transformations are applied?

  1. Lookup transformation
  2. Target transformation Correct answer.
  3. Router transformation
  4. Source transformation

Explanation

A Target transformation represents the destination object that receives the transformed rows at the end of the mapping flow. A Mapping must have at least one target, and its ports are mapped from upstream transformations. It defines the physical write object and its field structure.

Question 27Cloud Data Integration

Which task type is used to run a Mapping on a schedule and monitor each run's status in the Monitor service?

  1. Mass ingestion task
  2. Mapping task Correct answer.
  3. Synchronization task
  4. Replication task

Explanation

A Mapping task executes a Mapping and produces run instances that are tracked and monitored in the Monitor service. It supports scheduling, parameters, and runtime options. Synchronization, replication, and mass ingestion tasks serve different data movement patterns.

Question 28Cloud Data Integration

In a Mapping, what is the purpose of the Update Strategy transformation?

  1. To generate sequence numbers for new rows
  2. To sort rows before they reach the target
  3. To join two sources on a common key
  4. To flag each row as insert, update, delete, or reject for the target Correct answer.

Explanation

The Update Strategy transformation assigns a numeric constant to each row that tells the target how to treat it: insert, update, delete, or reject. This lets a single mapping perform mixed operations against a target table. Sorting, sequencing, and joining are handled by other transformations.

Question 29Cloud Data Integration

Which transformation passes rows through unchanged and is commonly used as a documentation or branching point in a mapping?

  1. Filter transformation
  2. Expression transformation Correct answer.
  3. Aggregator transformation
  4. Joiner transformation

Explanation

An Expression transformation can pass ports through unchanged while also calculating derived values, making it a common place to document logic or branch the flow. It evaluates expressions row by row. Filter, Aggregator, and Joiner transformations all change the row set or row count.

Question 30Cloud Data Integration

A developer wants to read a relational source and write to a flat file target in the same Mapping. Which two transformation types are required at minimum?

  1. A Router transformation and a Sorter transformation
  2. A Lookup transformation and an Aggregator transformation
  3. A Source transformation and a Target transformation Correct answer.
  4. Two Source transformations

Explanation

Every Mapping requires at least one Source transformation to read data and at least one Target transformation to write it. Intermediate transformations are optional and depend on the logic needed. The source and target objects are defined as part of the mapping design.

Question 31Cloud Data Integration

Which Secure Agent component is responsible for executing Mapping tasks and communicating with the Informatica Cloud services?

  1. The browser session used to open the web interface
  2. The target database client library alone
  3. The cloud-hosted metadata repository only
  4. The Secure Agent runtime installed on a host machine Correct answer.

Explanation

Mapping tasks run on the Secure Agent, a lightweight runtime installed on a machine that has network access to sources and targets. The agent registers with the cloud services and receives job instructions. The web interface and repository do not execute the mapping logic themselves.

Question 32Cloud Data Integration

When defining a flat file source in a Mapping, which property determines how columns are identified within each record?

  1. The tracing level of the task
  2. The target table's primary key
  3. The file format and delimiter settings Correct answer.
  4. The lookup cache size

Explanation

Flat file parsing depends on the file format definition, including delimiter, text qualifier, header handling, and column positions. These settings tell the agent how to split each record into fields. Target keys, lookup cache size, and tracing level do not affect source parsing.

Question 33Cloud Data Integration

Which statement describes how a Mapping task can be run with different source and target connections in different environments?

  1. By creating a new Secure Agent for each environment
  2. By changing the tracing level before each run
  3. By assigning connection values through task parameters or parameter files Correct answer.
  4. By editing the Mapping design for each environment

Explanation

Connection references in a Mapping task can be parameterized so the same mapping runs against different environments without redesign. Parameter files or task-level parameters supply the appropriate connection at runtime. Editing the mapping or adding agents is unnecessary for this purpose.

Question 34Cloud Data Integration

A Mapping must send rows that fail a validation rule to a separate target while valid rows continue to the main target. Which transformation supports this split in one pass?

  1. Aggregator transformation
  2. Sequence Generator transformation
  3. Router transformation Correct answer.
  4. Sorter transformation

Explanation

A Router transformation evaluates conditions and routes each row to the first matching output group, with a default group for rows that match none. Connecting the default group to an error target captures rejected rows while valid rows continue. Aggregator, Sequence Generator, and Sorter transformations do not split rows into conditional groups.

Question 35Cloud Data Integration

A developer needs to join a large transaction source with a small reference table and wants the reference data cached in memory. Which configuration is most appropriate?

  1. A Router transformation with one output group
  2. A Lookup transformation with a persistent or static cache on the small table Correct answer.
  3. An Aggregator transformation grouped by transaction ID
  4. A Joiner transformation with sorted input on both sources

Explanation

A Lookup transformation can cache a small reference table so each row is matched without repeated database calls. For a table that does not change during the run, a static or persistent cache reduces lookup overhead. A Joiner would require sorting and does not provide the same in-memory reference behavior.

Question 36Cloud Data Integration

A Mapping reads a source that contains duplicate customer records, and the developer wants to keep only one record per customer based on the most recent update timestamp. Which approach is most appropriate?

  1. Use a Router transformation with no conditions
  2. Use a Sequence Generator to renumber all rows
  3. Use an Update Strategy transformation to mark all rows as insert
  4. Sort by customer and timestamp, then use an Aggregator or Expression logic to keep the latest record Correct answer.

Explanation

Deduplication requires ordering the rows so the desired record per key can be identified. Sorting by customer and descending timestamp lets aggregation or expression logic retain the latest record per customer. Router, Sequence Generator, and Update Strategy do not by themselves remove duplicates.

Question 37Cloud Data Integration

A developer must add a sequence value that restarts at 1 for each new run and does not depend on the source data. Which transformation should be used?

  1. Aggregator transformation with a count port
  2. Lookup transformation with a dynamic cache
  3. Expression transformation with a variable port
  4. Sequence Generator transformation Correct answer.

Explanation

The Sequence Generator transformation produces sequential values and can be configured to reset for each session or continue across sessions. It generates values independently of source rows. Expression variables, lookups, and aggregations do not provide a managed sequence with restart behavior.

Question 38Cloud Data Integration

A Mapping task must run only after another Mapping task completes successfully. Which scheduling approach supports this dependency?

  1. Configure the dependent task to run on completion of the predecessor task Correct answer.
  2. Set the tracing level of the dependent task to Verbose
  3. Schedule both tasks at the same fixed time
  4. Run both tasks manually in parallel

Explanation

Task dependencies let a Mapping task be triggered by the successful completion of a predecessor task, ensuring correct ordering. Fixed simultaneous schedules and manual parallel runs do not guarantee ordering. Tracing level affects logging only.

Question 39Cloud Data Integration

In an Expression transformation, what is the effect of a variable port compared with an output port?

  1. A variable port writes its value directly to the target
  2. A variable port caches reference data across rows
  3. A variable port routes rows to a different output group
  4. A variable port stores a value that can be referenced by later ports in the same row processing Correct answer.

Explanation

Variable ports hold intermediate values that later ports, including output ports, can reference during row processing. They are evaluated in port order and are not written directly to the target. Caching reference data and routing rows are handled by other transformations.

Question 40Cloud Data Integration

A developer wants to filter source rows based on a condition that references a value supplied at runtime. Which configuration is most appropriate?

  1. A Filter transformation with a hard-coded constant condition
  2. A Sorter transformation with a runtime parameter
  3. A Filter transformation whose condition uses a mapping parameter Correct answer.
  4. A Sequence Generator with a runtime parameter

Explanation

A Filter transformation evaluates a condition per row, and using a mapping parameter in that condition lets the filter criteria change per run without editing the mapping. Hard-coded conditions cannot adapt at runtime. Sorter and Sequence Generator transformations do not filter rows.

Question 41Cloud Data Integration

A Mapping reads from a database source and writes to a database target. Which connection property must be configured for the Secure Agent to reach an on-premises database?

  1. A public cloud storage bucket
  2. A connection with network access from the agent host to the database Correct answer.
  3. A second metadata repository
  4. A browser extension installed on the developer's machine

Explanation

The Secure Agent executes the job, so its host must be able to reach the database over the network using the configured connection. Connection definitions supply host, port, credentials, and driver details. Browser extensions, cloud buckets, and extra repositories do not provide database connectivity for the agent.

Question 42Cloud Data Integration

A developer needs to load only rows whose values changed since the last run. Which approach is most appropriate?

  1. Reload all rows and mark them as insert
  2. Use a Router transformation to split rows by region
  3. Use a Sequence Generator to renumber all rows
  4. Use a source filter on the last modified timestamp and an Update Strategy or upsert target behavior Correct answer.

Explanation

Incremental loading uses a filter on a change indicator such as a last modified timestamp to select only new or changed rows. The target operation then inserts new rows and updates existing ones. Full reloads, renumbering, and region routing do not implement change detection.

Question 43Cloud Data Integration

A Mapping must aggregate sales by region and product, and the developer also wants a grand total across all groups. Which approach is most appropriate?

  1. Use a Joiner transformation on the same source twice
  2. Use a Filter transformation with no condition
  3. Use a single Aggregator with only group by ports
  4. Use one Aggregator for grouped totals and a second Aggregator without group by ports for the grand total, then combine the results Correct answer.

Explanation

Grouped aggregation and a grand total have different granularities, so separate Aggregator transformations are typically used and the results combined. An Aggregator with only group by ports produces grouped rows but no overall total. Filter and Joiner transformations do not compute aggregates.

Question 44Cloud Data Integration

A Mapping uses a Lookup transformation against a table that is updated earlier in the same mapping run. Which lookup cache configuration is required for the new rows to be visible to later lookups?

  1. A persistent cache with no refresh
  2. An uncached lookup with no connection
  3. A dynamic lookup cache that is updated as rows pass through Correct answer.
  4. A static lookup cache loaded once at the start

Explanation

A dynamic lookup cache is updated when new rows pass through the lookup, so subsequent rows in the same run can match them. A static cache is loaded once and does not see changes made during the run. A persistent cache without refresh and an uncached lookup without a connection cannot provide the required behavior.

Question 45Cloud Data Integration

A Mapping task fails because a target table rejects rows with null values in a required column. Which change best addresses the root cause?

  1. Increase the tracing level to Verbose
  2. Increase the lookup cache size
  3. Add validation or default-value logic upstream and route invalid rows to an error target Correct answer.
  4. Add a second Sequence Generator

Explanation

The failure is caused by data that violates the target constraint, so the mapping must handle or reject those rows before the write. Adding validation or default logic upstream prevents the constraint violation. Logging, cache size, and extra sequences do not fix invalid data.

Question 46Cloud Data Integration

A developer wants to reuse the same Mapping across development, test, and production with different source and target objects. Which design approach is most appropriate?

  1. Copy the Mapping three times with hard-coded objects
  2. Create three Secure Agents with identical settings
  3. Disable the target transformation in two environments
  4. Parameterize the source and target objects and supply values per environment Correct answer.

Explanation

Parameterizing source and target references lets one Mapping run against different objects per environment, reducing maintenance and drift. Duplicating the mapping creates parallel logic that must be kept in sync. Multiple agents and disabled targets do not provide environment-specific object selection.

Question 47Cloud Data Integration

A Mapping reads a very large source and performs a lookup against a reference table that changes frequently during the day. Which consideration most affects whether a static cache is appropriate?

  1. Whether the source file has a header row
  2. Whether the tracing level is set to Verbose
  3. Whether the reference data can change during the mapping run Correct answer.
  4. Whether the target table has a primary key

Explanation

A static cache is loaded once and does not reflect changes made during the run, so it is appropriate only when the reference data is stable for the duration of the job. Target keys, file headers, and tracing level do not determine cache validity. If the reference changes mid-run, a dynamic cache or uncached lookup may be needed.

Question 48Cloud Data Integration

A Mapping aggregates a large dataset and intermittently fails with out-of-memory errors on the Secure Agent. Which remediation is most appropriate?

  1. Set the tracing level to Verbose
  2. Increase the agent's available memory or reduce the data volume processed per run Correct answer.
  3. Add a second target transformation
  4. Remove the group by ports from the Aggregator

Explanation

Aggregation holds grouped data in memory, so out-of-memory errors indicate the agent lacks sufficient memory for the volume or the volume per run is too high. Increasing agent memory or processing less data per run addresses the cause. Verbose tracing, extra targets, and removing group by ports do not solve the memory constraint.

Question 49Cloud Data Integration

A developer must capture rows deleted from a source table between runs when the source has no soft-delete flag. Which approach is most appropriate?

  1. Compare the current source keys against the previously loaded keys and flag missing keys as deletes Correct answer.
  2. Reload all rows and mark them as insert
  3. Use a Sequence Generator to renumber all rows
  4. Use a Router transformation to split rows by region

Explanation

When the source does not flag deletes, deletions must be inferred by comparing current keys with the keys previously loaded. Keys present before but absent now are treated as deletes. Full reloads, renumbering, and region routing do not detect missing rows.

Question 50Cloud Data Integration

A Mapping joins two sources of different sizes, and the developer wants to minimize memory use on the agent. Which join approach is generally preferred?

  1. Use a Joiner transformation with both sources unsorted
  2. Use a Router transformation on both sources
  3. Use a Lookup transformation with the smaller source cached as the reference Correct answer.
  4. Use an Aggregator transformation on both sources

Explanation

Caching the smaller source as a lookup reference keeps memory use proportional to the small table rather than the large fact stream. A Joiner with unsorted inputs may require more memory for its work areas. Aggregator and Router transformations do not perform joins.