Sign in
Home/Oracle/Oracle GoldenGate 12c Implementation Essentials (1Z0-447)/Free questions

Oracle GoldenGate 12c Implementation Essentials (1Z0-447) — Free Practice Questions

10 free sample questions from a bank of 56, with the correct answers and explanations. No signup required — start practising right now.

1You must replicate to JMS Queues. Which product in the Oracle GoldenGate portfolio should be used?
  • Oracle Goldengate for Big Data
  • Oracle Goldengate Application Adapters
  • Oracle Goldengate Veridata
  • Oracle Data Integrator Adapters
Answer: B

B is correct. B -- Oracle GoldenGate Application Adapters. JMS queues are a message-oriented destination, and the Application Adapters product is the part of the portfolio that moves data into and out of application queues.

The short version

B -- Oracle GoldenGate Application Adapters. JMS queues are a message-oriented destination, and the Application Adapters product is the part of the portfolio that moves data into and out of application queues.

Key concepts

The GoldenGate portfolio is split by destination. GoldenGate for Big Data covers Kafka, Hadoop, NoSQL and flat-file style targets. Veridata is a data comparison and repair tool, not a replication path. Oracle Data Integrator is a separate ETL and data movement product with its own adapters. Application Adapters sit alongside GoldenGate to bridge application constructs such as JMS queues, user-defined files and messaging systems.

Why B are correct

B is correct because JMS is an application messaging construct rather than a database log, so replicating into a JMS queue needs the adapter layer. Application Adapters provide that: they are used with GoldenGate to deliver replicated data into JMS destinations, and the Oracle by-example that ships with the product is precisely a configuration that replicates from a Windows source into an ActiveMQ JMS destination. The other three products have no JMS capability at all.

Why the others are wrong

  • A. GoldenGate for Big Data targets Kafka, distributed file systems, NoSQL stores and other big-data endpoints. It has no JMS queue support, so it cannot be the product for this requirement.
  • C. Veridata compares two copies of the same data and repairs the differences. It does not replicate data into application queues and is not a data movement path at all.
  • D. Oracle Data Integrator is an independent ETL and data movement product. Its adapters connect to sources and targets for integration jobs, and they are not the GoldenGate component that writes replicated changes into JMS.

1z0-447 exam tip

Match the GoldenGate product to the destination type: databases and log-based capture for the classic engine, JMS and application constructs for Application Adapters, big-data stores for Big Data, and comparison for Veridata.

2Which statement is valid for DDL replication?
  • GoldenGate User Exits are not supported for DDL.
  • Source and target objects can be different.
  • Target objects can be different from source when using SOURCEDEFS in Replicat.
  • It can use WILDCARDRESOLVE IMMEDIATE.
Answer: C

C is correct. C -- target objects can differ from source objects when you use SOURCEDEFS in the Replicat.

The short version

C -- target objects can differ from source objects when you use SOURCEDEFS in the Replicat.

Key concepts

GoldenGate replicates DDL by parsing the statement and re-issuing it against the target. By default the object names in the DDL are used unchanged, so source and target must match. The Replicat's SOURCEDEFS parameter points at a file of source definitions, which is the documented mechanism for reconciling differing object names. On the Extract side, WILDCARDRESOLVE and COLMAP are DML features and have no meaning for DDL, while user exits are available to DDL through the UEXIT facility.

Why C are correct

C is correct because SOURCEDEFS exists precisely for the case where the target schema or object names differ from the source, and it is the only option in the list that describes a real, documented DDL-replication capability with that property. GoldenGate reads the source DDL and, using the sourcedefs, emits the equivalent statement against the correctly named target objects.

Why the others are wrong

  • A. User exits are supported for DDL. GoldenGate lets you embed a user exit call in a replicated DDL statement, so this claim is factually backwards.
  • B. Too loose to be the answer. DDL replication reproduces the statement with the source object names unless SOURCEDEFS is configured, so saying source and target objects simply can be different omits the one mechanism that actually makes it work and would be true of almost any statement.
  • D. WILDCARDRESOLVE is an Extract parameter for resolving wildcard table specifications in DML, typically resolving them to a single object. It is not part of DDL replication at all.

1z0-447 exam tip

For DDL, the mapping question is object naming. SOURCEDEFS on the Replicat is the answer whenever the target names differ; anything from the DML wildcard-resolution family is a distractor.

3You installed GoldenGate on your server, but you received an error when you tried to start GGSCI. What is the reason for this issue?
  • The incorrect version of GoldenGate was installed for your CPU architecture.
  • A license key must be applied first.
  • GoldenGate Manager was not started first.
  • The LD_LIBRARY_PATH variable to $ORACLE_HOME was not set.
Answer: A

A is correct. A -- the wrong build of GoldenGate was installed for the CPU architecture, so GGSCI cannot execute on that host.

The short version

A -- the wrong build of GoldenGate was installed for the CPU architecture, so GGSCI cannot execute on that host.

Key concepts

GGSCI is a compiled binary, so it has to match the machine it runs on. GoldenGate is published for both 32-bit and 64-bit platforms, and a build of the wrong width fails to load on the host's kernel rather than degrading gracefully. The same applies to the Extract, Replicat, Manager and logdump executables, because they are all built for the same platform as the product. On Linux the shared library path must also be right, but a wrong-architecture build fails earlier, before any library search can help.

Why A are correct

A is correct because the error occurs at launch, before any GoldenGate configuration is consulted, which points at the binary itself rather than at its environment or its licensing. A GoldenGate install built for the other CPU architecture cannot be executed on the host, and the reported failure follows immediately from ggsci not starting at all.

Why the others are wrong

  • B. GoldenGate is not licensed with a key that must be applied before the product runs. GGSCI is available as soon as the installation completes, which is exactly why the fault here has to lie with the binaries rather than with licensing.
  • C. Manager does not have to be running for GGSCI to start. GGSCI is the administration client and only contacts Manager when a command needs Manager to act, so a stopped Manager cannot prevent the client from launching.
  • D. This names the wrong variable and the wrong target directory. The shared library path that matters points at the GoldenGate installation directory, not at ORACLE_HOME, and in any case a bad library path produces a loader error naming the missing library rather than a bare failure to start.

1z0-447 exam tip

A GGSCI that will not start at all is a binary problem before it is a configuration problem. Check the build matches the host architecture, then check the library path.

4Identify the correct statement regarding Trails.
  • Extract and Pump are different OS level executable binary files.
  • Extracts cannot perform filtering, mapping, and transformations.
  • RMTTRAIL is for remote trails and EXTTRAIL is for local trails.
  • Extract is typically used to produce local trail files over a TCP/IP connection.
Answer: C

The short version

C — RMTTRAIL is for remote trails and EXTTRAIL is for local trails. It is the only true statement of the four.

Key concepts in this question

  • EXTTRAIL: parameter/command for a trail file residing on the local system.
  • RMTTRAIL: parameter/command for a trail file created on the remote (target) system.
  • Data pump: the secondary Extract that reads a local trail and ships it over TCP/IP to the remote trail.

Why C is correct

Oracle documentation defines the local trail versus remote trail exactly this way: a trail on the local system is an Extract (local) trail written via EXTTRAIL, while a trail on the target system is a remote trail written via RMTTRAIL, added on the source and linked to the data-pump Extract.

Why the others are wrong

  • A. Extract and data pump are the same OS executable; the process type comes from the parameter file (e.g. EXTTRAILSOURCE/RMTHOST/RMTTRAIL), not a different binary.
  • B. Extracts routinely perform filtering, mapping, and transformation via TABLE/MAP, WHERE, and conversion functions — that is core Extract work.
  • D. It is the data pump, not the primary Extract, that ships trail records over TCP/IP; a plain Extract produces local trail files.

1z0-447 exam tip

EXTTRAIL equals local, RMTTRAIL equals remote — and any option claiming Extract cannot transform data is always false.

5Replicat makes checkpoints in the trail file to mark its last read position. Select two ways to create checkpoint tables.
  • Run the chkpt_db_extracttrail.sql script.
  • Run the chkpt_db_create.sql script.
  • Use the ADD Extract Trail command.
  • Use the INFO REPLICAT command with the SHOWCH option.
  • Use the ADD CHECKPOINTTABLE command.
Answer: B, E

BE is correct. B and E -- run the chkpt_db_create.sql script, or issue ADD CHECKPOINTTABLE in GGSCI. Both create the checkpoint table.

The short version

B and E -- run the chkpt_db_create.sql script, or issue ADD CHECKPOINTTABLE in GGSCI. Both create the checkpoint table.

Key concepts

A Replicat records its read position so it can resume where it stopped. Oracle ships chkpt_db_create.sql to create the table directly, and ADD CHECKPOINTTABLE does the same thing through GGSCI so the table name is registered for future use. A checkpoint table is a global temporary table owned by the GoldenGate schema and is unrelated to the trail files themselves.

Why BE are correct

B is correct because chkpt_db_create.sql is the documented script for creating a checkpoint table directly in the target database. E is correct because ADD CHECKPOINTTABLE performs the same creation from GGSCI and additionally remembers the name so later ADD REPLICAT statements can inherit it. Both are genuine ways to satisfy the requirement, which is why the question asks for two.

Why the others are wrong

  • A. chkpt_db_extracttrail.sql is a control script for managing Extract trail files on disk, not a script that creates a checkpoint table in a database.
  • C. ADD EXTTRAIL is the GGSCI command that defines a trail file on disk. Checkpoints are stored in a database table, so a trail-file command cannot create one.
  • D. INFO REPLICAT ... , SHOWCH is a reporting command. It reads and displays checkpoint positions; it never creates the underlying table.

1z0-447 exam tip

Two routes to a checkpoint table: the chkpt_db_create.sql script, or ADD CHECKPOINTTABLE from GGSCI. Trail-file commands and INFO commands are about reading and writing trail files, never about creating tables.

6You want to generate an extract file or trail in a non-GoldenGate canonical format like those used in the Initial Load file-todatabase utility. How should you configure GoldenGate in order to accomplish this?
  • Configure Replicat with the FORMATASCII SQLLOADER OPTION.
  • Configure Extract with the FORMATSQL SQLLOADER OPTION.
  • Configure Extract with the FORMATASCII SQLLOADER OPTION.
  • Configure Replicat with the FORMATXML OPTION.
Answer: C

C is correct. C -- configure the Extract with FORMATASCII SQLLOADER. Only Extract writes the initial-load file, so only Extract chooses its format.

The short version

C -- configure the Extract with FORMATASCII SQLLOADER. Only Extract writes the initial-load file, so only Extract chooses its format.

Key concepts

During initial load the Extract group runs in a special mode and writes source records to an output file in a chosen format rather than to a trail. FORMATASCII SQLLOADER produces the delimited text layout that SQL*Loader expects, which lets a fast bulk loader consume the data. FORMATXML produces tag-delimited output for a user-defined format, and the trailing OPTION keyword selects which of those formats is written.

Why C are correct

C is correct because the file is produced by Extract, and the format keyword is an Extract parameter. FORMATASCII SQLLOADER makes Extract emit exactly the control-character-delimited layout that SQL*Loader's conventional path parses, so the two tools meet without a conversion step. The Replicat has nothing to do with writing the file at all.

Why the others are wrong

  • A. This puts a formatting parameter in the Replicat. The Replicat reads trail files and applies changes; it does not generate initial-load extract files and has no FORMATASCII SQLLOADER parameter.
  • B. FORMATSQL is not a GoldenGate format keyword. The valid forms are FORMATASCII, FORMATXML and FORMATBINARY, so this names a parameter that does not exist.
  • D. Same error as A, and an extra one: FORMATXML on the Replicat. Even if Replicat could format output, XML tag-delimited format is not what a SQL*Loader initial load consumes.

1z0-447 exam tip

File format is chosen by the writer, and during initial load the writer is the Extract. FORMATASCII SQLLOADER for SQL*Loader, FORMATXML for a user-defined parser -- both on Extract.

7Your Extract process, which is using an Oracle database, has stopped with uncommitted transactions. How should you guarantee an efficient recovery after Extract has stopped?
  • Use the FORCETRANS and PURGEORPHANEDTRANSACTIONS parameters to control the bounded recovery feature of the general Extract checkpointing facility.
  • Use the WARNLONGTRANS parameter to control the bounded recovery feature of the general Extract checkpointing facility.
  • Use the BR parameter to control the bounded recovery feature of the general Extract checkpointing facility.
  • Use the PURGEORPHANEDTRANSACTIONS parameter to control the bounded recovery feature of the general Extract checkpointing facility.
Answer: C

C is correct. C -- the BR parameter. Bounded recovery is what lets Extract restart cleanly with long-running open transactions still in flight.

The short version

C -- the BR parameter. Bounded recovery is what lets Extract restart cleanly with long-running open transactions still in flight.

Key concepts

Extract caches an entire transaction in memory from its first statement until it reaches the commit record, so that the trail always contains whole transactions. If the process stops while transactions are open, that cached data is written to disk and recovered on restart -- which is bounded recovery. The BR parameter controls it, with sub-parameters for start and stop age and for interval. BLOBMEMORY is a separate setting that caps how much memory large objects may consume before spilling.

Why C are correct

C is correct because bounded recovery is exactly the facility described: Extract stopped with open transactions, and the mechanism that guarantees a clean and efficient restart is BR. Tuning the bounded recovery behaviour with BR controls how old an open transaction may be before it is treated as bounded, and how often recovery data is examined, so recovery does not have to rescan the whole log.

Why the others are wrong

  • A. FORCETRANS and PURGEORPHANEDTRANSACTIONS do not control bounded recovery. FORCETRANS forces a transaction to be written to the trail even if it contains no captured data, and the other option relates to cleaning up records that were captured but never applied.
  • B. WARNLONGTRANS only controls the warning threshold for long-running transactions. It reports; it does not provide or tune any recovery mechanism.
  • D. PURGEORPHANEDTRANSACTIONS concerns discarding trail records that were extracted but never replicated, which is about trail hygiene rather than about restarting a stopped Extract.

1z0-447 exam tip

If the question is about restarting after a stop with transactions still open, the keyword is BR. BLOBMEMORY caps LOB memory, and WARNLONGTRANS only warns.

8Which database setting for the Oracle 11.2.0.4 database is required to enable Oracle GoldenGate?
  • ENABLE_GOLDENGATE_REPLICATION=true
  • ENABLE_STREAMS_REPLICATION=true
  • Configure the STREAMS_POOL_SIZE database parameter
  • ENABLE_DATABASE_REPLICATION=true
Answer: A

The answer is A, ENABLE_GOLDENGATE_REPLICATION=true. From 11.2.0.4 a database must be told that GoldenGate will replicate from it, and the flag lives in the compatible parameter, for example compatible='11.2.0.4'. It is a deliberate opt-in: without it Extract cannot register and replication fails before any trail is written.

9How should you configure Replicat to control how errors are handled in one-way replication?
  • Use REPERROR in the Replicat parameter file.
  • Use HANDLECOLLISION in the Replicat parameter file.
  • Use Conflict Detection and Resolution in the Replicat parameter file.
  • Use MAPEXCEPTION in the Replicat parameter file.
Answer: A

A is correct. A -- REPERROR in the Replicat parameter file. It is the parameter that decides what happens when an error is hit while applying a transaction.

The short version

A -- REPERROR in the Replicat parameter file. It is the parameter that decides what happens when an error is hit while applying a transaction.

Key concepts

Error handling in one-way replication is configured on the Replicat. REPERROR takes a number, IGNORE or REPERRORDELAY: a number skips the failed transaction that many times and then writes it to the discard file, IGNORE skips it indefinitely, and REPERRORDELAY defers it so later transactions can pass. HANDLECOLLISIONS is the older mechanism for the bidirectional case, and MAP/MAPERR write rejected records to a file.

Why A are correct

A is correct because REPERROR is the documented control for error handling in a one-way configuration. Setting it to a count, IGNORE or REPERRORDELAY tells the Replicat whether to abandon the transaction, keep retrying it, or set it aside, and any transaction that is finally abandoned lands in the discard file for later inspection.

Why the others are wrong

  • B. HANDLECOLLISIONS belongs to bidirectional replication, where a rollback caused by a conflicting row has to be retried automatically. In one-way replication there is no concurrent writer to collide with.
  • C. Conflict detection and resolution is the bidirectional mechanism. A one-way pipeline has exactly one writer per row, so there is no conflict to detect.
  • D. MAPEXCEPTION is not a GoldenGate Replicat parameter. Error routing is done with MAPERR to a file and REPERROR to control the behaviour.

1z0-447 exam tip

One-way error handling is REPERROR on the Replicat. HANDLECOLLISIONS and CDR are bidirectional features and do not belong in a one-way configuration.

10You must implement GoldenGate in a master-to-master configuration. What output from the profiling scripts should you look for to ensure that your implementation is correct?
  • Tables set with nologging
  • Tables with Large Object (LOB) data types
  • Tables that have conflicts
  • Triggers and sequences
Answer: C

C is correct. C -- tables that have conflicts. Conflict resolution only earns its keep in a bidirectional setup, and the profiling output tells you which tables are contended.

The short version

C -- tables that have conflicts. Conflict resolution only earns its keep in a bidirectional setup, and the profiling output tells you which tables are contended.

Key concepts

In a master-to-master topology both databases accept writes for the same tables, so the same row can be updated on both sides. GoldenGate detects this by comparing before and after images and resolves it with a configured rule. The profiling scripts that run before a bidirectional implementation are the tool that finds out which tables are actually shared, because a table with no overlap never needs a resolution rule.

Why C are correct

C is correct because a conflict is by definition a disagreement about the same row between two writers, which requires two masters. If the profiling output shows no conflicting tables, no conflict resolution configuration is needed and the implementation is simpler; if it shows conflicting tables, those are exactly the tables that need a rule.

Why the others are wrong

  • A. NOLOGGING tables are a performance and recoverability concern because their changes are not written to the redo log in the normal way. It says nothing about two writers contending for a row.
  • B. LOB columns are a capacity and ordering concern. GoldenGate handles them, and their presence neither creates nor resolves a conflict.
  • D. Triggers and sequences are classic sources of conflicts and integrity problems in a bidirectional setup, but they are not what the profiling output is being searched for here; the question asks what indicates that resolution is required.

1z0-447 exam tip

In a bidirectional build, the profiling step exists to answer one question: which tables have two writers? Only those tables need a conflict resolution rule.

Want the full bank of 56 questions for Oracle GoldenGate 12c Implementation Essentials (1Z0-447)? See all practice exams.