What is the Qlik Replicate certification?

The Qlik Replicate certification is Qlik’s credential for practitioners who design and run data replication and change data capture pipelines on the Qlik Replicate platform. It carries the exam code QREP, runs to 50 questions across four weighted domains, and is aimed at someone already operating the product rather than someone evaluating it.

Qlik Replicate moves data between systems continuously rather than in scheduled batches. It reads changes at the source, usually from a transaction log, and applies them at the target, which is why the exam’s vocabulary is endpoints, tasks, transformations and exceptions rather than jobs and schedules.

What makes this credential unusually easy to plan for is that Qlik publishes the domain weightings itself. Most vendors leave that to third parties or omit it entirely. Here the vendor and the money site agree on all four figures, which removes the guesswork that normally dominates the first week of study.

Why does one domain carry nearly half the marks?

Design holds 47 percent of the QREP exam, Administration 23 percent, Troubleshooting 22 percent and Operations 8 percent. On a 50 question paper that is roughly 24 questions in Design and about four in Operations, so the four domains are not four equal parts of a syllabus but one dominant subject with three supporting ones.

The reason for the imbalance is in the nature of the product. Replication is a configuration discipline. Once a task is designed correctly it runs without intervention, which is exactly why day-to-day operation is worth only four questions while the decisions that produced the task are worth twenty four.

That distribution also tells you what kind of questions to expect. A domain worth nearly half the paper cannot be tested by asking where a button is, so Design questions tend to give a requirement and ask which configuration satisfies it. The sample sets on the money site’s QREP practice exam show that shape clearly, which is useful for calibrating early rather than late.

How is the QREP exam delivered and scored?

QREP is 50 questions in 120 minutes with a 54 percent pass mark, priced at $250 USD and registered through Qlik. Qlik publishes the question count, duration, pass mark and all four domain weightings on its own exam-details page, and every figure matches the money site’s syllabus exactly.

Two of those numbers point in opposite directions and both are worth planning around. A 54 percent pass mark is low, meaning 27 correct answers out of 50 and a margin of 23 questions. That is the most generous allowance of any exam covered here.

The time is generous too. Fifty questions in 120 minutes is 144 seconds each, roughly double what a typical vendor product exam allows. Taken together, a low bar and a long clock usually signal one thing: the questions are expected to take thinking rather than recall. That matches a paper whose largest domain asks which configuration meets a requirement.

What sits inside the Design domain?

Design has four objectives and they escalate. Endpoints, then architecture, then task type and task settings, then transformations. Each one is a decision the person configuring a replication pipeline has to make before anything runs, and together they account for roughly 24 of the 50 questions.

Endpoints and architecture

Endpoints are the source and target definitions, and the objective is specifically about the requirements to create and manage them. Different source systems expose change data differently, which is why the endpoint is not a connection string but a set of decisions about how changes will be read.

The architecture objective sits underneath that. Understanding how the platform captures and applies changes is what makes the endpoint choices comprehensible rather than arbitrary. Qlik’s Replicate introduction documentation is the clearest statement of that model and is worth reading before anything else in this domain.

The underlying mechanism is not proprietary. Reading committed changes out of a database’s own write-ahead log is how most change data capture works, and the PostgreSQL logical replication documentation explains the same idea from the database side. Understanding it there makes the endpoint objective substantially easier, because you stop treating each source system as a special case.

Task type and transformations

The third objective asks you to determine which task type to use and which settings go with it, from a stated requirement. That is the heart of the exam. A full load, an ongoing change capture, or a combination of the two are different task types with different consequences for latency, load and recovery, and the requirement in the question is what decides between them.

The fourth objective does the same for transformations: given a requirement, determine which transformation to use. Transformations in a replication pipeline are deliberately limited compared with a full integration tool, so the skill being tested is knowing what belongs in the pipeline and what does not. A relational target and an event streaming target behave very differently once changes start arriving, which is exactly the kind of consequence the task settings objective expects you to have met.

Administration is almost a quarter of the paper

Administration carries 23 percent, roughly 12 questions, across four objectives: server settings, matching user types to roles, setting up the Enterprise Manager, and listing the deployment options. This is the platform side rather than the pipeline side, and it is where candidates who have only ever configured tasks tend to be thin.

Enterprise Manager deserves particular attention because it is a separate component rather than a screen. It is the layer that gives a single view across multiple Replicate servers, which means it exists for estates rather than for single installations. Anyone who has only worked with one server may never have opened it.

The user types and roles objective is phrased as a matching exercise, which is a hint about question style. Expect items that name a responsibility and ask which role grants it, rather than items that ask you to describe the permission model in the abstract.

Deployment options round it out. Where the product runs, and in what configuration, is a question asked at purchase time and rarely revisited, so it is another objective that experience alone may not have covered.

What does the troubleshooting domain actually ask for?

Troubleshooting is 22 percent, about 11 questions, and it is unusually specific for a domain of that size. Four objectives: getting logs from a task, setting error handling and debug log levels, obtaining a diagnostic package, and debugging errors using the attrep_apply_exceptions table.

That last objective is the one worth memorising by name. The exceptions table is where the platform records rows it could not apply at the target, and it is the first place an experienced operator looks when a task is running but data is missing. A question naming it is testing whether you have actually debugged a live pipeline.

The diagnostic package objective is similarly practical. It is the bundle you generate when escalating to support, and knowing what it contains and when to produce it is the difference between a resolved ticket and a week of back and forth.

Error handling settings sit alongside debug logging because they answer the same question from two directions: what should the task do when a row fails, and how much detail should it record while deciding. Twenty two percent is a large share for diagnostics, and it reflects a product whose failures are usually silent rather than loud.

Why is Operations only 8 percent?

Operations carries 8 percent, about four questions, and just two objectives: determining when and why to start or stop a task, and identifying the options for handling task metadata. It is the smallest domain on the syllabus by a wide margin, and that is the most revealing number on the page.

A well designed replication pipeline does not need operating. It runs, and the only genuine operational decisions are when to interrupt it and what happens to its state when you do. Stopping a change capture task is not like stopping a batch job, because the source keeps generating changes while you are stopped, so the decision has consequences that a scheduled job does not have.

Task metadata is the other half of that. What the platform remembers about where a task had reached determines whether restarting resumes or reloads, and that single behaviour is why the objective exists at all.

The planning lesson from an 8 percent domain is not to skip it, since four questions is more than a seventh of the 23 question margin. It is to give it an evening rather than a week, and to spend the time saved in Design.

Does Qlik expect you to have used the product first?

Yes, explicitly. Qlik recommends a minimum of one year of practical experience with the Replicate platform before sitting the exam, alongside expertise in onboarding sources and an understanding of the platform’s components. That is a firmer readiness statement than most vendors publish and it is not on the money site’s syllabus page.

A year is a meaningful bar, and the domain weights explain why. An exam that spends nearly half its marks asking which configuration satisfies a requirement is difficult to pass from documentation, because the answers depend on consequences you have seen rather than rules you have read.

There is no prerequisite certification, so nothing blocks an earlier attempt. The recommendation is about the likelihood of passing rather than about eligibility, and Qlik states it on its Qlik Replicate exam details page alongside the weightings.

For anyone mapping out a wider Qlik profile, the vendor’s credentials differ sharply in how much they publish. The Qlik AI Specialist certification is structured quite differently from this one, which is worth knowing before assuming a single preparation approach covers the portfolio.

How should you prepare for QREP?

Preparation for QREP should follow the weightings rather than the domain order. Roughly half your time belongs in Design, a quarter split between Administration and Troubleshooting each, and a single evening in Operations. The low pass mark means the goal is competence across all four rather than mastery of any one.

1. Read the Replicate architecture documentation first, because every Design objective assumes you know how changes are captured at the source and applied at the target.
2. Build endpoints against at least two different source systems, so that the differences in how each exposes change data become concrete rather than theoretical.
3. Create a full load task and a change capture task against the same endpoint pair, and watch what each does, since the task type objective is the single densest part of the exam.
4. Apply every transformation type the product offers to a running task, so that determining which transformation a requirement calls for becomes recognition rather than reasoning.
5. Set up Enterprise Manager even if you only have one server, because it is a separate component that task-level experience never introduces you to.
6. Break a task deliberately by forcing rows to fail at the target, then read the attrep_apply_exceptions table and generate a diagnostic package from the result.
7. Stop and restart a change capture task and observe what the retained metadata does to the resume, which is essentially the whole Operations domain in one exercise.
8. Finish with timed sets of 50 questions, using the full 120 minutes rather than rushing, since the generous clock is there because the questions expect reasoning.

Four to six weeks is realistic for someone already running Replicate in production. Someone with less than Qlik’s recommended year should expect the Design domain to be the constraint, and should spend the extra time building pipelines rather than reading about them.

Practitioners coming from the data quality side of the Qlik portfolio will find the vocabulary familiar but the emphasis different, and the Talend Data Quality Implementer exam is a useful reference point for how differently Qlik weights its data credentials.

Frequently Asked Questions

How many questions are on the QREP exam?

Fifty questions in 120 minutes, which is about 144 seconds each. That is roughly double the per-question time most vendor product exams allow.

What is the passing score for the Qlik Replicate certification?

Fifty four percent, meaning 27 correct answers out of 50 and a margin of 23. Qlik publishes the same figure on its own exam-details page as the money site’s syllabus does.

How much does QREP cost?

$250 USD, registered through Qlik. The price is published on the money site’s syllabus page rather than on Qlik’s exam-details page, which directs candidates to a purchase link instead.

Which QREP domain carries the most marks?

Design, at 47 percent, which is roughly 24 of the 50 questions. Administration follows at 23 percent, Troubleshooting at 22 percent and Operations at 8 percent.

Does Qlik publish the domain weightings itself?

Yes, and all four match the money site’s syllabus exactly. That is unusual. Most vendors publish topic names without percentages, which makes planning a guess rather than a calculation.

How much experience does Qlik recommend before the exam?

A minimum of one year of practical experience with the Replicate platform, plus expertise in onboarding sources and an understanding of the platform’s components.

Is there a prerequisite certification?

None. The one year experience recommendation is guidance on readiness rather than an eligibility condition, so nothing prevents an earlier attempt.

What is the attrep_apply_exceptions table?

It is where the platform records rows it could not apply at the target, and the syllabus names it directly under troubleshooting. It is the first place to look when a task is running but data is missing.

Why is the Operations domain so small?

Because a correctly designed replication task largely runs itself. The only real operational decisions are when to stop or start a task and what happens to its retained metadata, which is exactly what the two objectives cover.

Does the exam cover Enterprise Manager?

Yes, under Administration. Enterprise Manager is a separate component that gives a view across multiple Replicate servers, so anyone who has only worked with a single installation may never have opened it.

Conclusion

QREP is a design exam with three supporting domains attached: 50 questions, 120 minutes, $250, and a 54 percent bar that is the lowest of any credential covered here. Design alone is 47 percent, Administration 23, Troubleshooting 22 and Operations 8, and Qlik publishes every one of those figures itself.

Plan to the weighting rather than the list. Spend half your preparation building and comparing tasks, give Administration and Troubleshooting a genuine block each, and let Operations have a single evening. Then use the full two hours in the exam, because a low pass mark and a long clock together mean the questions were written to be thought about.

What Is the Oracle GoldenGate 26ai Certification?

1Z0-948 leads to the Oracle GoldenGate 26ai Implementation Associate credential. It tests the installation, configuration, parameterisation and monitoring of Oracle’s real-time data replication platform across seven topics, from architecture components through mapping and transformation to migration. The paper runs 50 questions in 90 minutes and requires 68 percent to pass. Replication targets frequently sit on autonomous infrastructure, which the 1Z0-931-26 exam guide covers in detail.

Oracle positions GoldenGate as a replication and integration platform that moves committed changes between heterogeneous systems in near real time, and the GoldenGate product overview frames it that way. The associate credential validates the person who builds and runs those pipelines rather than the architect who designs the target estate.

What makes the exam distinctive within Oracle’s catalogue is how component-specific it is. Most associate-level Oracle exams reward broad familiarity. This one asks which process writes to a local trail and which reads from a remote one, and there is no way to reason your way to that from first principles. Replication is one route to availability and clustering is another, which is the ground the Oracle RAC certification covers.

What Does 26ai Actually Refer To?

26ai is the GoldenGate release the exam is built against, and it is the current one. The credential is titled Oracle GoldenGate 26ai Implementation Associate, and the older exam in the same line, 1Z0-947, covers GoldenGate 19c. If you are choosing between them, 1Z0-948 is the newer paper and the one aligned to the release most organisations are moving toward. Replication is one route into Oracle data movement; Oracle EPM data integration is the other, and its exam is weighted very differently.

The practical difference the version makes is architectural. The syllabus names the Distribution Service and the Receiver Service as separate components, and those belong to the microservices architecture rather than to the older classic deployment model. That single detail tells you which documentation set to revise from.

Oracle’s GoldenGate documentation set is organised around that architecture, which makes it the natural companion to the syllabus. Working from older classic-architecture material is the most common way candidates end up learning component names that the exam no longer uses.

Why Should You Learn the Architecture Topic First?

Because six of the seven topics reference components that topic two introduces. Topic two asks you to describe GoldenGate components, create Capture process types for Oracle database, create Replicat process types, explain the Extract, Distribution Service, Receiver Service, Replicat and local and remote trails, and configure the process recovery mechanism. Everything after it assumes you know those names.

The five names that matter most

Extract reads change records from the source. Trails are the files those records land in, and the distinction between local and remote is which side of the network they sit on. The Distribution Service moves trail data outward, the Receiver Service accepts it at the far end, and Replicat applies it to the target. That is the whole path, and almost every scenario in the exam sits somewhere on it.

Process recovery is the objective people skip

The last objective in topic two is configuring the process recovery mechanism, and it is easy to read past. It is also the thing that decides whether a pipeline survives a restart without losing or duplicating rows, which makes it exactly the sort of objective a 50-question paper likes to test once and decisively.

Learn this topic properly and topics three through seven become recognition rather than memorisation. Learn it last and you will read the other six twice.

How Much of 1Z0-948 Is Installation and Initial Load?

Topic three, Installing and Configuring Oracle GoldenGate, carries six objectives and is the largest installation block on the paper. It covers downloading and installing the product, synchronising source and target with the Initial Load, preparing the database for change data capture, configuring replication component parameter files, using the command interface to create processes, and using the GoldenGate utilities.

Two of those deserve more attention than they usually get. The Initial Load is how a replication pair starts life in sync, and getting it wrong means every subsequent change is applied to a target that was never correct to begin with. Preparing the database for change data capture is the source-side work that makes any of it possible, and it is prerequisite knowledge for topics four and six as well.

• Initial Load is a synchronisation step, not an installation step, and the syllabus deliberately places it here
• Change data capture preparation is source-database work, so it sits outside the GoldenGate binaries entirely
• The command interface objective is about creating processes, so expect questions phrased as tasks rather than as syntax recall
• The utilities objective is broad and under-specified, which makes hands-on exposure the only reliable preparation

Data replication of this kind is a recognised discipline in its own right, and the data integration knowledge area of the DAMA body of knowledge gives useful vocabulary for why initial synchronisation and ongoing capture are treated as separate problems.

What Do the Parameter Topics Expect You to Know?

Topic four, Defining Oracle GoldenGate Parameters, has three objectives: describing GLOBALS parameters, creating solutions using component parameters for replication requirements, and installing GoldenGate parameters. The distinction between global and component-level settings is the whole topic, and it is the sort of hierarchy question that is easy to get backwards under time pressure.

GLOBALS parameters apply across the deployment. Component parameters apply to a single Extract or Replicat. The middle objective is the interesting one because it is phrased as creating solutions for replication requirements, which means the questions are likely to describe a requirement and ask which parameter delivers it, rather than asking what a named parameter does.

Configuration options sit in a separate topic

Topic six, Configuring Oracle GoldenGate, covers DDL replication, compression and encryption options, event actions based on use cases, and Integrated Capture, Replicat and deployment options. Read alongside topic four, this is really the second half of the parameter story: topic four is where settings live, topic six is what the significant ones do.

DDL replication is worth singling out. Replicating data is one problem and replicating schema changes is another, and candidates who have only run steady-state pipelines often have no exposure to the second at all.

How Does the Exam Treat Mapping and Transformation?

Topic five is the smallest on the paper with just two objectives: implementing use cases for transformation functions, and implementing macros for transformation. Small does not mean skippable, because with seven topics across 50 questions even a two-objective topic can carry several marks.

Transformation functions handle the row-level work of reshaping a column between source and target. Macros are the reuse mechanism, letting a piece of parameter logic be written once and referenced repeatedly. The exam frames both as implementation rather than description, which again points at scenario-shaped questions.

The practical advice here is narrow and useful. Write one macro. Not five, not a library, just one that does something real in a parameter file, and then read it back. The mechanics stop being abstract immediately, and the syllabus asks for nothing deeper than that.

What Does Managing, Monitoring and Migrating Cover?

Topic seven bundles four objectives that share an operational character: managing OGG command and data security, managing GoldenGate command and data security, implementing and troubleshooting GoldenGate Monitoring, and configuring the migration process. It is the topic that describes life after the pipeline is running.

Security appears twice in the topic’s own wording, which is unusual and worth taking at face value. Command security governs who can issue operational instructions, and data security governs what moves across the wire and how it is protected. Both are examinable and both are frequently left to defaults in real deployments, which is precisely why candidates find them unfamiliar.

Monitoring is stated as implementing and troubleshooting, not merely configuring. Expect at least one question that describes a symptom rather than asking for a setting. Migration closes the topic and closes the syllabus, covering the move from one deployment to another, which in practice usually means moving off the classic architecture.

What Does 1Z0-948 Cost and What Score Do You Need?

1Z0-948 costs USD $245 and requires 68 percent to pass, which is 34 correct answers out of 50. The paper runs 90 minutes, giving an average of 108 seconds per question. Oracle notes that pricing may vary by country or by localized currency, so confirm the figure for your region when booking.

No weightings are published for this syllabus, which removes the usual planning shortcut. The nearest available proxy is objective count: topic three has six objectives, topic two has five, topics six and seven have three and four, topic four has three, and topics one and five have two each. Weight your time roughly that way and adjust for what you already know.

Treat that as a guide and nothing firmer. Seven topics across 50 questions averages seven questions per topic, and a two-objective topic can still produce a question about a macro you have never written. You can see the raw 1Z0-948 topic list if you want to do the objective count yourself.

Who Is the Implementation Associate Level For?

The credential fits database administrators, data engineers and integration specialists who build and operate replication pipelines. Associate level here means implementation rather than design: you are being asked whether you can stand a pipeline up and keep it running, not whether you can architect a multi-region topology from scratch.

The clearest self-test is the component vocabulary. If Extract, trail, Distribution Service and Replicat are words you use rather than words you have read, the exam validates work you already do. If they are new, you are learning a substantial platform in order to sit an associate paper about it, which only makes sense if GoldenGate is arriving in your environment anyway.

On career value, replication and integration skills sit at the higher end of database administration because the work is unforgiving and the failure modes are visible to the business. Current benchmarks for database administrator pay give a sensible baseline before adding a replication speciality on top.

If you are coming from the previous generation of the product, the site’s GoldenGate 12c exam guide is a useful comparison point for how much the architecture has moved, and the wider Oracle credential index shows where the associate level sits in the rest of the catalogue.

How Should You Sequence Preparation for Seven Unweighted Topics?

Sequence beats allocation when there are no weightings to allocate against. Start with the architecture topic so the component names stop being abstract, then install something so the architecture stops being theoretical, then work outward through parameters, transformation and operations in the order the syllabus itself implies.

1. Start with the Overview of the Oracle GoldenGate Architecture, learning Extract, the Distribution Service, the Receiver Service, Replicat and the difference between local and remote trails until you can draw the path without prompting.
2. Add the process recovery mechanism to that same session, because it is the objective most candidates read past and the one that decides whether a restarted pipeline stays correct.
3. Install GoldenGate on a real pair of databases next, running an Initial Load end to end so that synchronisation and change data capture preparation become something you have done rather than something you have read.
4. Move to Defining Oracle GoldenGate Parameters, fixing the distinction between GLOBALS parameters and component parameters firmly before anything else in that topic.
5. Take Configuring Oracle GoldenGate immediately afterwards, since DDL replication, compression, encryption and Integrated Capture are the settings that give the parameter hierarchy something to do.
6. Write one transformation macro in a real parameter file and read it back, which covers the mapping and transformation topic more effectively than any amount of reading about it.
7. Finish with Managing, Monitoring and Migrating, paying particular attention to command security and data security, both of which are named twice in the topic and left at defaults in most real deployments.
8. Sit timed papers at 50 questions in 90 minutes, keeping any single question under two minutes so an unfamiliar utility cannot cost you the marks you had already earned.

One caution on sources. Oracle’s certification pages are frequently unavailable to anything but a signed-in browser, so anchor your preparation to the product documentation and a running deployment rather than to a catalogue page you may not be able to open on the day you need it.

Frequently Asked Questions

How many questions are on the 1Z0-948 exam?

The exam contains 50 questions and runs for 90 minutes, which is an average of 108 seconds per question. That is comfortable for a recall item and tight for a scenario that describes a broken replication path.

What score do you need to pass 1Z0-948?

68 percent, which means 34 correct answers out of 50. With seven topics and no published weightings, no single topic is small enough to write off and still reach that mark reliably.

How much does the Oracle GoldenGate 26ai certification cost?

USD $245. Oracle notes that pricing may vary by country or by localized currency, so check your regional price before booking rather than assuming the dollar figure applies everywhere.

What is the difference between 1Z0-948 and 1Z0-947?

1Z0-948 is the GoldenGate 26ai Implementation Associate exam and 1Z0-947 covers GoldenGate 19c. If you are choosing between them, 1Z0-948 is the newer paper and aligns to the current release.

Which topic has the most objectives?

Installing and Configuring Oracle GoldenGate, with six: download and install, Initial Load, change data capture preparation, replication component parameter files, the command interface, and the GoldenGate utilities.

Does the syllabus publish domain weightings?

No. It lists seven topics with their objectives and assigns no percentages. Objective count is the nearest usable proxy, but treat it as a rough guide rather than a blueprint.

What are local and remote trails?

Trails are the files that captured change records are written into. Local trails sit on the source side and remote trails on the target side, and the Distribution and Receiver Services are what move data between them.

Do you need hands-on GoldenGate experience to pass?

It is very difficult without it. The objectives are phrased as create, configure and implement rather than describe, and questions about the utilities and the command interface reward exposure over reading.

Is DDL replication on the exam?

Yes. It appears in the Configuring Oracle GoldenGate topic alongside compression and encryption options. Candidates who have only run steady-state data pipelines often have no exposure to schema change replication.

Who should take this certification?

Database administrators, data engineers and integration specialists who build and run replication pipelines. It validates implementation ability rather than architectural design, which is what associate level means here.

Conclusion

1Z0-948 rewards precision about components far more than it rewards general knowledge of replication. Extract, trails, the Distribution and Receiver Services and Replicat form a single path, and almost every question sits somewhere along it. Learn that path first, install the product and run an Initial Load so the path becomes real, then work outward through the parameter hierarchy, one transformation macro, and the operational topic that closes the syllabus. There are no weightings to plan against, so use objective count as a rough guide and let hands-on exposure fill the gaps. Read the topic list carefully before you book, and make sure the component names are ones you use rather than ones you recognise.