What does COH284 actually ask you to prove?

COH284 verifies that you can configure, manage, administer and troubleshoot a NetBackup environment day to day. Cohesity’s own description names basic NetBackup architecture, configuring backups and restores, configuring storage devices, implementing specialised backup solutions, and troubleshooting failed jobs, device and media connectivity problems and disaster recovery situations including catalog recovery.

Read that list again and notice what is missing. There is no design domain, no capacity planning, no architecture selection. This is an operator’s credential, aimed at the person who keeps an existing estate running rather than the person who specified it.

The product history explains a lot about the credential’s shape. NetBackup is decades old and arrived at Cohesity through the Veritas data protection merger, which is why the vendor name on the certificate is newer than the software, and why the official exam page still sits at a URL carrying the old Veritas exam code. If your NetBackup knowledge predates the rebrand, it has not gone stale.

What are the COH284 exam facts?

COH284 is 60 questions in 105 minutes with a 65 percent pass mark, priced at 200 USD and registered through Cohesity rather than a third-party test centre network. That allows 105 seconds per question, which is generous by certification standards and reflects how much scenario reading the paper contains.

One note on sourcing. The duration, fee, pass mark, language, expiry and retake rule all appear on Cohesity’s own exam page and match the exam catalogue exactly. The question count of 60 does not appear in the vendor’s published list, so treat it as the catalogue figure rather than a vendor-confirmed one.

Because the paper is weighted toward running an estate rather than building one, the fastest way to find your real gaps is to answer items in the exam’s own shape, and a COH284 practice test will show whether your monitoring knowledge matches your configuration knowledge.

Why does monitoring outweigh configuring?

Monitor and Maintain NetBackup is worth 36 percent, the single largest domain, while Configure NetBackup is worth 31 percent, Troubleshoot NetBackup 23 percent and Tune NetBackup 10 percent. Nearly three fifths of the paper therefore lands on what happens after the estate is built, which mirrors how the job actually divides.

The Monitor and Maintain objectives are unusually specific. They name image management through the administration console, verifying, expiring, importing and manually duplicating images, managing disk and cloud storage, interpreting reports, and deciding when to prioritise, cancel, suspend, resume, restart, retry or manually run a job. They also name role-based access control, host ID certificates, applying updates through VxUpdate, and configuring anomaly and malware detection.

That last pair is the clearest signal of where backup software has moved. Detecting ransomware behaviour inside backup data is now an administration task rather than a security team’s problem, and it sits inside the largest domain of an operator exam. The control frameworks agree: data recovery is a top-level control in the CIS Controls precisely because recovery is what an attack tests.

The storage and deduplication objectives that catch people out

The Configure domain reaches much further than policies and schedules. It names synthetic backups, True Image Restore, multiple data streams, checkpoint restart, disk staging, Storage Lifecycle Policies, Auto Image Replication, Isolated Recovery Environment and NetBackup Accelerator, and then a full deduplication list on top of that.

The deduplication objectives are the ones candidates underestimate. Media server deduplication, client-side deduplication, optimised duplication, KMS options, MSDP cloud storage, WORM storage, Universal Shares, Snapshot Manager and storage servers all appear by name. These are not interchangeable features, and questions turn on which one solves a stated constraint rather than on what each one is.

A worked example makes the pattern obvious. If a remote site has a thin link and a short backup window, client-side deduplication moves less data across it; if the requirement is instead that copies cannot be altered for a retention period, WORM storage is the answer and deduplication is irrelevant. The exam asks that kind of question repeatedly, which is why memorising feature definitions produces a near miss rather than a pass.

Why immutability keeps appearing

WORM storage, KMS-managed encryption and an isolated recovery environment are three objectives pointing at one idea: a backup copy an attacker cannot quietly modify. Federal storage security guidance sets out the same requirement in detail, and NIST SP 800-209 is the reference worth reading once if these controls are new to you, because it explains why they exist rather than merely how to switch them on.

What does a catalog recovery question look like?

Troubleshoot NetBackup is 23 percent of the paper and the objectives name failed jobs, device and media connectivity problems, common disaster recovery situations and catalog recovery specifically. A catalog question tends to describe a loss, give you a partial set of facts, and ask what must be true before a restore can even begin.

The catalog is where the pattern is clearest, because it is the one component whose loss changes the order of everything else. If the catalog is gone, the images may still exist on storage but the system cannot find them, so the recovery sequence starts with the catalog backup rather than with the data, and a candidate who has never rehearsed that sequence will pick a plausible wrong answer.

Tune NetBackup, at 10 percent, is the smallest domain and the vaguest: analyse, optimise and tune. In practice that means recognising where a bottleneck sits when a job runs slowly rather than fails, which is a different diagnostic habit from fault finding and worth six questions.

The vendor terms almost nobody else publishes

Cohesity publishes four commercial terms on its exam page that most third-party write-ups omit entirely: there are no prerequisite exams or certifications, the exam is delivered in English, the credential expires after two years, and a failed attempt can be retaken once every 14 days.

The two-year expiry is the one to plan around. A credential with a two-year life is a commitment to revisit the product on a fixed cycle, and for a working NetBackup administrator that is usually reasonable, since the platform changes in that time anyway. For someone certifying to clear a hiring filter rather than to do the work, it is a recurring cost worth knowing about before booking.

The 14-day retake window is mild by industry standards and effectively removes the case for gambling on an early attempt: waiting a fortnight is rarely worse than sitting underprepared, and the second attempt costs the full fee again. Both facts come straight from the official exam page, which is worth checking yourself before you book in case the terms move.

Where does this sit in the Cohesity certification track?

Protection Professional is the working administrator tier of the Cohesity scheme, below the Architect Expert credential and alongside the Security Specialist track. Because it carries no prerequisites, it can be a first Cohesity credential or a later one, and the NetBackup variant is distinct from the Cohesity-platform variant that shares the tier.

That distinction causes real confusion. COH-285 is the Protection Professional exam for the Cohesity data platform itself, while COH284 is the NetBackup one, and search demand for the two codes runs side by side. They are not alternative names for one exam, and material written for one does not prepare you for the other.

If security controls rather than day-to-day operations are your direction, the specialist route is the better next step, and the COH350 security specialist credential goes deep on the areas COH284 touches only through its anomaly detection and RBAC objectives.

Who should sit this exam?

COH284 suits backup administrators running NetBackup daily, infrastructure engineers who inherited it as part of a wider platform, managed service staff supporting multiple NetBackup estates, and Cohesity partners who need a demonstrable product credential. Cohesity recommends roughly three months of hands-on experience, and that is a fair floor rather than a formality.

It is a weak fit for two groups. Anyone who has never operated NetBackup will find the monitoring and troubleshooting domains, 59 percent of the paper between them, almost impossible to prepare for from reading alone. Anyone looking for a vendor-neutral data protection credential will find this deeply product specific, which is its value if you run the product and its limitation if you do not.

For someone earlier in the Cohesity track, the platform-side entry credential is the gentler introduction, and the COH125 career path route makes more sense before taking on an exam weighted this heavily toward operational recall.

How should you prepare for 60 questions in 105 minutes?

Prepare against the objective list, weight your time toward the two operational domains, and rehearse the sequences rather than the definitions. With 105 seconds per question the clock is not the enemy, so the real risk is answering plausibly rather than correctly on scenarios you have never actually performed.

1. Read the published objectives and mark every feature you have configured yourself. The unmarked ones are your study list.
2. Spend roughly 60 percent of your time on Monitor and Maintain and Troubleshoot. They are 35 of the 60 questions.
3. Rehearse a catalog recovery end to end in a lab, because the sequence is what the questions test.
4. Build a one-page table of the deduplication options with the constraint each one solves.
5. Work through the job-state verbs deliberately: prioritise, cancel, suspend, resume, restart, retry and manually run are not synonyms, and the exam treats them as distinct.
6. Sit a full timed set before booking, and treat anything below 70 percent as a signal to go back to the objectives rather than to the calendar.

Anomaly and malware detection deserves a dedicated evening. It is newer than most of the syllabus, it sits in the largest domain, and it is the area where experienced NetBackup administrators are most likely to be out of date.

Frequently Asked Questions

How many questions are on the COH284 exam?

Sixty questions in 105 minutes, which allows about 105 seconds each.

What is the passing score for COH284?

Sixty-five percent, which on a 60 question paper means 39 correct answers.

How much does the Cohesity NetBackup certification cost?

Two hundred US dollars, registered through Cohesity rather than a third-party test centre network.

Does COH284 have prerequisites?

None. Cohesity states plainly that there are no prerequisite exams or certifications, though it recommends about three months of practical experience.

How long is the credential valid?

Two years. Cohesity publishes the expiry on the exam page, so plan for a recertification cycle rather than a one-off.

What happens if I fail COH284?

You may retake it after 14 days, at the full fee. There is no published discount for a second attempt.

What is the difference between COH284 and COH-285?

COH284 is the Protection Professional exam for NetBackup; COH-285 is the Protection Professional exam for the Cohesity data platform. Different products, different objectives, and material for one does not prepare you for the other.

Which domain carries the most marks?

Monitor and Maintain NetBackup at 36 percent, ahead of Configure at 31 percent, Troubleshoot at 23 percent and Tune at 10 percent.

Is this still a Veritas certification?

No. NetBackup came to Cohesity through the Veritas data protection merger, so the credential is issued by Cohesity even though the product and much of the older study material carry Veritas branding.

Is the exam available in languages other than English?

Cohesity lists English only on the exam page.

Conclusion

COH284 is an operator’s credential for people who run NetBackup rather than design it. Sixty questions, 105 minutes, 65 percent to pass, 200 USD, no prerequisites, English only, and a two-year life.

The weighting is the whole story. Monitoring, maintaining and troubleshooting carry 59 percent of the marks between them, so preparation that concentrates on configuration steps will feel productive and score badly. Rehearse the sequences instead: a catalog recovery, a stalled duplication, a job that has to be suspended rather than cancelled.

Read Cohesity’s published terms before booking, give anomaly and malware detection an evening of its own, and sit a timed set first. Fourteen days between attempts is a long time to wait for something a fortnight of targeted preparation would have fixed.

What Is the Nutanix Multicloud Infrastructure Certification?

NCP-MCI validates that an administrator can deploy, manage, protect and monitor a Nutanix environment that spans more than one cloud. It runs 75 questions in 120 minutes, costs $200 USD, and is scheduled through Nutanix directly rather than a third-party test provider. Nutanix classifies it at the advanced level of its certification programme.

The word multicloud in the name is not decoration. Every one of the four domains is phrased around a multicloud environment, and the objectives reflect that: cloud integration, cross-cluster VM migration, replication between sites, and capacity planning that has to hold up when one location fails. An administrator who only ever touches a single on-premises cluster will find parts of this exam describe work they have never done.

Registration and the current course catalogue both live behind the Nutanix University portal, which requires a sign-in, so plan on creating an account before you try to book anything. A consolidated view of the credential and its objectives sits on the NCP-MCI certification page.

Why Is NCP-MCI Scored Out of 6000?

NCP-MCI reports a scaled score between 1000 and 6000, with 3000 required to pass. Scaled scoring exists so that different forms of the same exam can be compared fairly: a harder set of questions is worth slightly more per correct answer than an easier one, and the reported number accounts for that before you ever see it.

The practical consequence is that you cannot convert 3000 into a number of correct answers, and you should not try. Candidates who go in planning to “get sixty out of seventy-five” are working from a model the exam does not use. What you can rely on is that every question contributes, and that no domain is discounted.

What this changes about preparation

With no published percentage weights across the four domains and a scaled result at the end, triage is not available. The usual tactic of writing off the smallest domain has nothing to aim at here. The efficient approach is breadth: make sure you can answer something sensible in all four areas rather than mastering two and hoping.

Should You Take NCA First, or Go Straight to NCP-MCI?

There is no formal requirement to hold the Nutanix Certified Associate credential before sitting NCP-MCI, so the decision is purely practical. If you already administer a Nutanix cluster day to day, go straight to the professional exam. If your exposure is limited to demos, evaluations or a lab you built once, the associate credential builds the vocabulary far more cheaply than a failed professional attempt does. The automation track sits alongside it, and the NCP-MCA certification guide covers what Self-Service and X-Play add on top.

The honest test is whether the following sound like descriptions of your week or like unfamiliar phrases: Prism Central deployment sizing, protection domains, LCM upgrades, vDisk live migration, and Cluster Lockdown Mode. If four of the five are familiar, book NCP-MCI.

Where each credential lands

NCA proves you understand the platform. NCP-MCI proves you can run it, including when a site fails. Employers reading a CV treat them very differently, and the professional credential is the one attached to Nutanix’s advanced tier. For candidates starting further back, this Nutanix Certified Associate walkthrough covers what the entry exam involves.

Whichever route you take, the full programme structure is published on the Nutanix certifications page.

Which Version of NCP-MCI Should You Study?

Version confusion is the single biggest avoidable risk on this exam. NCP-MCI has been revised repeatedly as the platform has moved, and material for retired versions is still widely circulated and still ranks in search results. Studying the wrong one produces confident answers that are wrong for the current form.

The blueprint has moved with the product in ways that are easy to spot once you know to look. Prism Central has taken on far more of the management surface, which is why Prism Central deployment, sizing, scaling and its Policy Engine now appear as first-class objectives rather than as an afterthought. Prism Central based disaster recovery sits alongside the older Prism Element based approach rather than replacing it, and the exam expects you to know when each applies.

How to avoid the trap

• Check the version listed on the current objective set before buying or trusting any study material.
• Treat anything that only mentions protection domains, and never mentions Prism Central based DR, as out of date.
• Verify version-sensitive behaviour against your own deployment, and note where your cluster is older than the exam expects.
• Prefer resources that name the current version explicitly over those that just say “NCP-MCI”.

A current list of preparation material for this exam is collected in this NCP-MCI preparation resource guide.

What Does the Cluster Management Domain Cover?

The cluster domain is the foundation of the exam and the broadest of the four. It spans deploying and configuring clusters, standing up and sizing Prism Central, network management, cluster storage, and cluster security, and it expects configuration-level knowledge rather than concepts.

Deployment questions get specific quickly. You are expected to gather the required configuration details, tell deployment methods apart, size the controller VM according to which features are enabled, configure the required services, and perform node imaging. Prism Central is treated as its own deployment exercise, with sizing driven by which features you intend to switch on and a scaling path when the estate grows.

Networking and storage inside the domain

Networking covers virtual switches, virtual subnets, VPC configuration, and Flow load balancing and traffic mirroring. Storage covers what to do when a cluster runs short, when to apply which resiliency model, when reserved capacity or reserved space is appropriate, and how external storage integrates.

Security is heavier than most candidates expect

Cluster security is a substantial sub-area in its own right: hardening through nCLI, disabling SSH and Bash access, filesystem whitelists, SSL certificate implementation, firewall requirements, Cluster Lockdown Mode, encryption options, network segmentation, and identity and access management with role-based access control in Prism Central. Anyone treating this exam as purely a virtualisation paper will be surprised here.

How Deep Does the Workload Management Domain Go?

The workload domain moves from the platform to what runs on it, and it goes deeper into day-to-day operations than the cluster domain does into architecture. Virtual machine management, storage containers, workload storage, VM conversion and workload security all sit here, and all are examined as tasks.

Migration is the recurring theme. The objectives call out moving VMs both within a cluster and between clusters, live migration of vDisks, and full conversion of workloads from other platforms, including installing VirtIO, identifying supported source and target platforms, and choosing the right Nutanix tool for the job. Anyone who has run a real migration will recognise the shape of these questions immediately.

Workload security

Security appears again, this time at the workload level: enabling vTPM and Secure Boot on a VM, applying VM level encryption through a storage policy, applying microsegmentation policy, analysing an environment with Security Central, and creating approval policies. Microsegmentation in particular rewards hands-on time, because the mental model of policy applied per workload rather than per network segment takes a while to settle.

The underlying architecture that makes all of this possible is standard hyperconverged infrastructure, and candidates coming from traditional three-tier environments benefit from getting that model straight before anything else.

What Do the DR and Operations Domains Ask For?

These two domains together cover what happens when the environment is under stress. Disaster recovery and data protection asks how you protect and recover Prism Central itself, how you configure Prism Central based DR for VMs and volumes, and how the older Prism Element based approach still works. Operations and monitoring covers capacity, performance and maintenance.

Disaster recovery

Protecting Prism Central is listed first, which is telling: if the management plane is gone, recovering the workloads becomes much harder. The objectives name continuous and point-in-time backup, instance migration, the recovery process itself, and what has to happen after a restore. Replication choices then follow, spanning asynchronous and near-sync, synchronous replication and Metro, and container-level disaster recovery.

The Prism Element side has not gone away. Protection domains across multiple clusters, network mapping for failover, remote site configuration, planned and unplanned failover, and Metro with ESXi are all still examinable, which is why studying only the newer approach leaves a gap.

Operations and monitoring

Capacity management is the standout here because it is forward looking: predicting growth, understanding limits and risks, configuring proactive alerting, optimising resources, and validating that capacity still holds under a failure or DR scenario. Performance work covers Storage QoS, automated resolution of performance alerts, VM performance analysis, and reading audit and system logs. Maintenance covers expanding a cluster, ejecting nodes, diagnosing NCC errors, and planning and executing LCM upgrades.

Which Roles Does This Certification Serve?

NCP-MCI serves infrastructure roles where someone is accountable for uptime rather than for a project. Virtualisation administrators, systems administrators running Nutanix estates, infrastructure engineers, and the operations side of platform teams all draw on what the credential proves, and it maps closely onto a job description rather than onto a technology list.

The credential also carries unusually strong salary signalling. Nutanix records that Skillsoft ranked NCP-MCI the third highest paying IT certification in its published list, dated November 2024. That is a vendor-reported figure rather than an independent one, so treat it as a signal rather than a promise, and check it against real market data such as independent systems administrator pay data for your own region.

Where it does not help

It is not a design credential and it is not an automation credential. Nutanix runs separate professional exams for multicloud automation, database automation, unified storage, end-user computing, cloud native and network security, and an expert-level track above them. If your work is mostly building automation on top of Nutanix rather than running the platform, one of those is the better fit.

Frequently Asked Questions

How many questions are on the NCP-MCI exam?

The exam has 75 questions and a 120 minute limit, which allows roughly 96 seconds per question. That is comfortable for recall items but tighter than it looks once scenario questions with several plausible answers appear.

What score do you need to pass NCP-MCI?

You need 3000 on a scale that runs from 1000 to 6000. Because the score is scaled rather than a raw percentage, it cannot be converted into a fixed number of correct answers.

How much does the Nutanix multicloud infrastructure certification cost?

The exam fee is $200 USD. Training such as the Enterprise Cloud Administration course is priced separately, so budget for that on top if you plan to take the recommended path.

Do you need NCA before taking NCP-MCI?

No formal prerequisite exists, so you may register directly. In practice the associate credential is worth taking first if your Nutanix exposure is limited to labs and evaluations rather than daily administration.

Where do you book the NCP-MCI exam?

Through Nutanix rather than a third-party test provider, using the Nutanix University portal. You will need to create an account there before you can schedule, so allow time for that step.

How many domains does NCP-MCI have?

Four: managing clusters, managing workloads, disaster recovery and data protection, and operations and monitoring. No percentage weights are published for them, so none can safely be skipped.

Does the exam cover security?

Substantially. Cluster security covers hardening, lockdown mode, certificates, encryption and network segmentation, while workload security adds vTPM, Secure Boot, VM level encryption and microsegmentation policy.

Is Prism Central heavily tested?

Yes. Prism Central deployment, sizing, scaling and its Policy Engine are named objectives, and protecting and recovering Prism Central itself opens the disaster recovery domain, which is a strong signal of its importance.

How long does preparation usually take?

Two to three months is realistic for an administrator already running Nutanix, and considerably longer without daily exposure. The disaster recovery material is the usual bottleneck because it is hardest to practise safely.

What comes after NCP-MCI?

Nutanix runs further professional exams in multicloud automation, database automation, unified storage, end-user computing, cloud native and network security, plus an expert-level multicloud infrastructure credential above them.

Conclusion

NCP-MCI rewards administrators who run Nutanix rather than those who have read about it. Four unweighted domains, a scaled result at the midpoint of a 1000 to 6000 range, and objectives that reach from node imaging through microsegmentation to Metro failover leave very little room to specialise your way through the paper.

The two decisions worth getting right before you book are both outside the blueprint. Confirm which version the current objective set describes, because material for retired versions still circulates freely and reads perfectly plausibly. And be honest about whether the associate credential should come first, using daily familiarity with Prism Central, protection domains and LCM upgrades as the test rather than ambition.