Buyer’s Guide - August 2026

The best NVMe and SSD health monitoring tools, ranked

Q: Can I monitor NVMe SSD health from Linux?

Yes. smartctl from smartmontools reads NVMe health information directly, and nvme-cli provides the nvme smart-log command for direct health reads. The kernel also exposes NVMe temperatures via hwmon automatically. For monitoring platforms, Netdata has a dedicated NVMe collector that wraps nvme-cli via ndsudo, and both the Zabbix SMART template and Prometheus smartctl_exporter work on Linux hosts.

Disk performance charts will not warn you that a drive is dying. Health monitoring reads SMART and NVMe attributes - estimated endurance, available spare, media errors, critical warnings - and that is a different job from tracking IOPS. We ranked nine tools on health metric depth, fleet coverage, alerting, and what it actually costs to run them at scale.

Start with Netdata (free) Jump to rankings

Why this list exists

Most teams discover NVMe health monitoring the hard way: a drive hits 100% of its rated endurance or throws a critical warning, and the monitoring stack that faithfully tracked capacity and latency said nothing. That is because health and performance are different data. Performance comes from the block layer; health comes from the NVMe SMART log, and most general-purpose platforms never read it.

The mistake buyers make is assuming their existing disk-usage or IO dashboards cover health. They usually do not. Before shortlisting anything here, check three dimensions against your fleet:

Attribute depth. Does the tool read NVMe-native fields - percentage used, available spare, composite temperature, media and data integrity errors, the critical warning bitfield - or only generic ATA SMART attributes? Tools built for SATA often show incomplete data on NVMe drives.
Collection cadence. A daily or 6-hour health poll is an audit, not a monitor. Thermal events and media errors move in minutes. Decide whether you need 10-to-60-second polling or whether scheduled audits plus alerting are enough.
Alerting out of the box. Some tools ship triggers for endurance exhaustion and critical warnings; others hand you the metrics and leave you to write PromQL or plugin thresholds yourself.

We deliberately do not quote list prices in this guide. Pricing pages change, and per-sensor, per-service, and per-node models are not comparable as raw numbers anyway. Instead, each card describes the pricing shape - what the bill scales with - and links the vendor’s official pricing page. For hands-on reference material, our operator guides for NVMe monitoring cover the attributes and commands in depth.

We assembled the shortlist from vendor documentation, collector source repositories, and practitioner threads where engineers report what actually caught a failing drive. Desktop utilities were included alongside fleet platforms because searchers for this topic compare both, and the ranking makes the distinction explicit.

Health metric depth carries the most weight because it is the whole point of the category: a tool that only reads temperature and capacity is not monitoring NVMe health. Fleet scalability and alerting follow, since a single-host utility cannot protect a storage fleet no matter how good its readouts are. Deployment cost matters because the pricing model (per-node, per-service, per-sensor, per-series) determines whether monitoring every drive stays affordable as the fleet grows.

Tester credit

Compiled by the Netdata team - Updated August 12, 2026

Sources verified

Scoring criteria

NVMe/SSD health metric depth 25%
NVMe-native attributes vs generic SMART fields
Fleet scalability and centralized view 20%
Single pane across many hosts vs per-machine checks
Alerting and failure prediction 15%
Built-in triggers vs write-your-own rules
Deployment and operational cost 15%
Infrastructure you operate and how the bill scales
Historical trends and visualization 10%
Retained SMART history in dashboards vs CLI output
Ecosystem and integrations 10%
Fit with Prometheus, Nagios, Grafana, Windows tooling
Ease of setup 5%
Time from install to useful health visibility

Vendor 01 / 09 · #netdata

Netdata

Real-time infrastructure monitoring with dedicated NVMe and SMART collectors, per-second dashboards, and built-in anomaly detection.

Best for

Fleet operators who want NVMe health, SMART attributes, and disk performance in one per-second dashboard
Teams that want alerting and anomaly detection without standing up a separate Prometheus and Grafana stack
Hybrid fleets mixing Linux and BSD hosts with NVMe devices

Pricing

Per-node subscription; Netdata Cloud Business starts at $4.5/node/month on annual plans, with the per-node price decreasing as node count grows
Free Cloud tier for small fleets, capped by connected node count
Agents are open source (AGPL); you run and operate them
Bill grows with the number of concurrently running agents; containers on a monitored host are included at no extra cost

Pros

Dedicated NVMe collector reads estimated endurance, available spare, composite temperature, IO transferred, power cycles, critical warnings, unsafe shutdowns, media errors, and thermal management transitions via nvme-cli
Separate SMART collector (smartctl) covers non-NVMe drives with error rates, power-on time, temperature, and vendor attributes
NVMe collector defaults to a 10-second interval and is tunable; general collection is per-second
Built-in alert on the NVMe critical warnings state, plus anomaly detection via Netdata AI
800+ integrations with an open-source agent
Unlimited metrics, logs, users, and retention on paid plans; no per-GB or per-metric charges

Where teams pair it

The NVMe collector needs nvme-cli installed and privileged access (ndsudo) on each host, and the SMART collector does not auto-detect devices - you configure device selectors or extra devices
Only the NVMe critical-warnings alert ships built in; the SMART collector has no default alerts, so you define your own thresholds
UI-based configuration of the NVMe collector requires a paid Netdata Cloud plan; on the free tier you edit collector configs on the host

Verdict

Netdata is the only tool in this list with both a dedicated NVMe collector and a SMART collector feeding the same per-second dashboards, alerting, and anomaly detection, with no separate time-series database to run. That combination is why it ranks first for this topic: you get endurance, spare capacity, temperature, media errors, and critical warnings at 10-second granularity, next to the IO performance metrics that give them context. Per-node pricing with unlimited metrics and retention also avoids the per-sensor and per-GB traps that penalize drive-dense hosts. The honest caveat: you need nvme-cli and privileged access on every host, and SMART device selection is manual. Teams with heavy custom-alerting requirements on non-NVMe drives sometimes pair it with a scripted smartd layer.

Start free ->

Vendor 02 / 09 · #zabbix

Zabbix

Open-source enterprise monitoring with an official SMART template that discovers and monitors NVMe, SSD, and HDD health.

Best for

Enterprises that want a self-hosted platform with no per-node license and a maintained SMART template
Teams already running Zabbix who need NVMe health alongside network and server monitoring
Organizations that want vendor support SLAs without per-device pricing

Pricing

Open source (AGPL), no license fee; you run and operate it
Paid support subscriptions priced by response coverage, support contacts, and number of Zabbix servers and proxies covered
Zabbix Cloud is a managed SaaS option with its own monthly pricing
Bill grows with support level and covered servers, not per monitored host

Pros

Official ‘SMART by Zabbix agent 2’ template discovers HDD, SSD, and NVMe disks and reads temperature, percentage used, critical warning, media errors, and power-on hours
Built-in triggers for NVMe endurance over 90%, high temperature, failed self-tests, and smartctl exit-status bits
No external scripts required; uses smartctl via Zabbix agent 2 with a sudoers entry
Unlimited hosts, metrics, and alerts in the open-source product
Active and passive agent modes plus native high-availability options

Cons

SMART template items refresh on a 6-hour interval for most attributes, so this is a health audit, not a real-time view
Requires Zabbix agent 2 and smartmontools 7.1+ on every host, plus sudo privileges for smartctl
The NVMe self-test item prototype is not discovered for NVMe disks in the official template
No built-in anomaly detection; alerting is threshold-based

Verdict

Zabbix has the most complete packaged SMART support of the open-source platforms: discovery, NVMe-aware items, and ready-made triggers for endurance over 90% and high temperature. For an existing Zabbix shop, adding NVMe health is straightforward. The tradeoff is cadence: a 6-hour refresh means a thermal runaway or a burst of media errors can age out before you see it. Treat Zabbix as the audit and alerting layer for slow-moving health trends, and pair it with something faster if your failure mode is thermal.

Read full Netdata vs Zabbix -> Zabbix pricing ->

Vendor 03 / 09 · #prometheus

Prometheus + smartctl_exporter

Open-source metrics stack that scrapes SMART and NVMe health data from every host via the smartctl exporter.

Best for

SRE and platform teams already standardized on Prometheus
Kubernetes-centric environments where exporters deploy as DaemonSets
Organizations that want to build custom NVMe health dashboards in Grafana

Pricing

Open source, self-hosted; you run and operate Prometheus, exporters, and storage
Optional Grafana Cloud is usage-based: per active series for metrics, per GB for logs and traces, with a free tier capped by series count and retention
Bill grows with active series count and data point rate, plus any managed Grafana usage

Pros

smartctl_exporter (prometheus-community) wraps smartctl JSON output and exposes NVMe and SATA SMART metrics on a /metrics endpoint
Default 60-second scrape interval with a 10-minute device rescan
Grafana dashboard 22604 provides ready-made panels for temperature, media errors, critical warnings, wear leveling, and device lifetime
Works with any Prometheus-compatible storage (Mimir, VictoriaMetrics, Thanos)
Active open-source community and no vendor lock-in

Cons

You assemble and operate Prometheus, exporters, Alertmanager, and Grafana yourself
smartctl_exporter needs privileged access to block devices, typically a privileged container or root
No NVMe health alerting rules ship with the exporter; you write your own PromQL
NVMe drives do not expose ATA-style vendor attributes, so some SMART fields are simply unavailable

Verdict

This is the standard fleet approach for teams already living in Prometheus: deploy the exporter, import dashboard 22604, and you have 60-second NVMe health polling across every node. The metric coverage is solid because smartctl reads the NVMe health log directly. What you are buying with your time is the assembly: no alerting rules ship with the exporter, so endurance and critical-warning alerts are your PromQL to write and maintain. If your organization already has that muscle, this is excellent. If not, the hidden cost is the operational stack around the metrics.

Read full Netdata vs Prometheus -> Grafana pricing ->

Vendor 04 / 09 · #checkmk

Checkmk

Agent-based infrastructure monitoring with a SMART plugin that tracks disk health statistics and error counters.

Best for

Teams that want a packaged monitoring appliance with agent auto-discovery
Organizations that prefer per-service pricing over per-host or per-GB models
Mixed environments needing SMART health alongside application and network monitoring

Pricing

Free tier (Community/Free mode) capped by service count and a single site
Commercial editions (Pro, Ultimate, Cloud) priced per monitored service, with custom metrics and synthetic tests as additional units
Bill grows with the number of services; a SMART check on each disk counts as a service
30-day trials for all commercial editions

Pros

The smart_stats check monitors error counters and temperature via the smart_posix agent plugin
Agent-based discovery inventories disks and creates per-disk services automatically
Per-service pricing with no per-host fees
Scales to 100,000+ hosts in self-hosted editions
Alerting, dashboards, and reporting included in one product

Cons

The SMART check is primarily documented for HDDs reporting Temperature_Celsius; NVMe attribute depth (endurance, spare, media errors) is thinner than in Netdata or Zabbix
A check goes critical when counters increase and only returns to green after re-inventory, which creates manual toil
The free tier is capped by service count, limiting large fleets without a paid subscription

Verdict

Checkmk is a strong general-purpose platform, and its per-service pricing is refreshingly predictable for teams burned by per-GB bills. For SMART health specifically, though, it treats disks as one check type among thousands, and the NVMe-native attribute set is not modeled as deeply as in the leaders here. Drive-dense hosts also multiply your service count, which is exactly where the bill grows. A good pick if Checkmk is already your standard; a shallower one if NVMe health is the primary job.

Read full Netdata vs Checkmk -> Checkmk pricing ->

Vendor 05 / 09 · #smartmontools

smartmontools

The open-source foundation for SMART monitoring: smartctl for inspection and smartd for background health tracking and alerts.

Best for

Sysadmins who want a lightweight, scriptable health layer under a larger monitoring stack
Heterogeneous fleets with mixed SATA, SAS, and NVMe drives
Users who want scheduled self-tests and email alerts without a full monitoring platform

Pricing

Open source, self-hosted; you run and operate it
No license fee; packaged in most Linux distributions and available for Windows, macOS, and BSD
Operating cost is your own infrastructure and configuration effort

Pros

Industry-standard SMART tooling since 2002; supports ATA/SATA, SCSI/SAS, and NVMe drives
smartd runs continuously, schedules short and long self-tests, and emails alerts on threshold breaches
NVMe support includes health status, error log, temperature, percentage used, and media errors
Integrates with Nagios, Zabbix, Prometheus, and other tools via smartctl output
Runs on Linux, Windows, macOS, and BSD

Cons

No built-in web dashboard or historical graphing; output is CLI and log based
NVMe drives do not expose ATA-style vendor attributes, so some SMART fields are unavailable
smartd.conf configuration is manual and per-device
Email alerting requires a working local mail setup

Verdict

Almost every tool on this page wraps smartmontools, which tells you what it is: the engine, not the car. On a single host or a small fleet, smartd with scheduled self-tests and email alerts is a legitimate health monitoring setup with no license fee beyond your infrastructure and configuration effort. What you give up is history, visualization, and any fleet-level view. Most teams end up here anyway, because their platform’s collector calls smartctl underneath. Knowing this layer well makes you better at running whatever sits above it.

Visit smartmontools -> smartmontools docs ->

Vendor 06 / 09 · #scrutiny

Scrutiny

A purpose-built SMART health dashboard that merges drive metrics with real-world failure rates from Backblaze.

Best for

Homelab and self-hosted users who want a polished SMART web UI quickly
Fleet owners who want failure-rate context from Backblaze data to set thresholds
Users already running Docker who want a single-container SMART dashboard

Pricing

Open source (MIT), self-hosted; you run and operate it
All-in-one Docker image, or manual install with separate collector and web containers
Uses InfluxDB for metric storage, which you also operate

Pros

Web UI focused on critical SMART metrics with historical trends
Merges manufacturer SMART metrics with real-world failure rates from Backblaze, which grounds thresholds in observed data
Auto-detects connected drives via smartctl –scan, including NVMe (with SYS_ADMIN capability in Docker)
Configurable alerting via webhooks
MIT-licensed with an active community and maintained Docker images

Cons

Collector runs on a daily cron schedule by default, so health data is not real-time
Requires InfluxDB for storage, adding operational overhead
The project describes itself as a work-in-progress; some features are still planned
No built-in paging or escalation; alerting stops at webhooks

Verdict

Scrutiny is the most purpose-built SMART dashboard in open source, and the Backblaze failure-rate merge is a genuinely smart idea: it replaces arbitrary thresholds with observed field failure data. For a homelab or a small storage fleet, the all-in-one container gets you a useful health view in minutes. The limits are cadence and operations: daily collection will not catch a thermal event, and you own the InfluxDB underneath. A reporting tool for drive health, not a real-time monitor.

Scrutiny on GitHub -> Docker image ->

Vendor 07 / 09 · #prtg

PRTG Network Monitor

Windows-centric network monitoring with WMI-based disk health sensors for physical and virtual drives.

Best for

Windows-centric IT teams that want an all-in-one monitoring console
Small-to-mid-size organizations that prefer a GUI over YAML and PromQL
MSPs monitoring many customer sites from a single PRTG instance

Pricing

Subscription priced per sensor, billed annually
Freeware edition capped by sensor count, roughly a small office deployment
Bill grows with the number of sensors; a disk health check on each drive consumes sensors
30-day trial with unrestricted sensors

Pros

WMI HDD Health sensor monitors SMART attributes on Windows drives
WMI Disk Health sensor (added 2024) monitors physical and virtual disk status on Windows servers
250+ native sensor types covering network, server, and storage in one console
Freeware tier for small environments
Agentless WMI/SSH/SNMP collection reduces per-host installs

Cons

NVMe-specific attributes (endurance, available spare, media errors) are not first-class sensors; coverage depends on WMI/SMART passthrough
Per-sensor licensing makes drive-dense fleets expensive; every disk health check consumes sensors
Windows-centric; Linux monitoring runs over SSH and is less deep
Threshold-based alerting only; no anomaly detection

Verdict

PRTG earns its place for Windows shops that want disk health inside a broader monitoring console without touching a config file. The WMI path is convenient and agentless. But convenience is the whole offer: NVMe-native health fields are not modeled as sensors, and per-sensor licensing means a host with eight NVMe drives spends eight-plus sensors on health alone. For a virtualization-heavy Windows estate that is often acceptable. For a Linux NVMe fleet, look elsewhere on this list.

Read full Netdata vs PRTG -> PRTG pricing ->

Vendor 08 / 09 · #nagios

Nagios + check_smart

The classic open-source monitoring platform, extended with the check_smart plugin for HDD, SSD, and NVMe health checks.

Best for

Long-time Nagios shops that want to add SMART checks to existing hosts
Teams comfortable writing NRPE commands and plugin configurations
Environments where a lightweight plugin is preferred over a full agent

Pricing

Nagios Core is open source (GPL), self-hosted; you run and operate it
Nagios XI is a separate paid commercial product
Exchange plugins such as check_smart are community-maintained
Bill grows with the commercial license and the infrastructure you operate

Pros

The check_smart plugin (Napsty fork) monitors HDD, SSD, and NVMe drives via smartctl
Supports drives behind hardware RAID controllers (MegaRAID, HP CCISS, Intel RAID)
Per-attribute threshold tuning for fine-grained alerting
Runs as a standard plugin, so it works with Nagios, Icinga, and compatible forks
Active community maintenance and documentation

Cons

Requires root or sudo access to smartctl on every monitored host
No built-in dashboard for SMART trends; you get status and alert history only
Plugin configuration is manual per host and per drive
Nagios Core has no native time-series storage; historical SMART data requires add-ons

Verdict

If Nagios already runs your infrastructure, check_smart is the pragmatic way to add NVMe health checks without introducing a new platform, and the RAID-controller support is a genuine strength most collectors lack. What it cannot give you is trend data: Nagios tells you a drive crossed a threshold, not that its media error count has doubled every week for a month. That blind spot matters for failure prediction, which is the point of health monitoring. Fine as an alerting bolt-on; insufficient as the whole strategy.

Read full Netdata vs Nagios -> Nagios pricing ->

Vendor 09 / 09 · #crystaldiskinfo

CrystalDiskInfo

A Windows desktop utility that reads SMART data from HDDs, SSDs, and NVMe drives and shows health status in a simple interface.

Best for

Individual workstations and small offices that want a quick SMART health readout
Technicians diagnosing a single drive before replacement
Windows users who want a lightweight resident SMART monitor with email alerts

Pricing

Freeware (proprietary), donation-supported
No license fee; runs on Windows XP through Windows 11 and Windows Server
Operating cost is per-workstation installation and manual review

Pros

Reads SMART data from HDD, SSD, and NVMe drives with a clear Good/Caution/Bad health status
Resident mode with alarm and email notification features
Supports Intel RAID and some USB drives alongside direct-attached NVMe and SATA
Lightweight, portable, and widely recommended in practitioner forums
Active development with regular releases

Cons

Windows-only; no Linux, macOS, or server fleet support
No centralized dashboard or multi-host view; each machine is checked individually
No API or scriptable interface for integration with other monitoring tools
Health status derives from SMART thresholds, which some practitioners find unreliable for NVMe

Verdict

CrystalDiskInfo is the right answer to a different question: is this one drive, on this one Windows machine, healthy? For that it is excellent, fast, and carries no license fee. It appears in this ranking because searchers compare it against fleet platforms, and the distinction deserves to be explicit: there is no fleet view, no API, no history, and no server story. Keep it in the technician’s toolkit. Do not mistake it for monitoring.

Visit CrystalDiskInfo -> Tom's Hardware SSD health guide ->

Frequently asked questions

What SMART attributes matter most for NVMe SSD health?

Start with percentage used, which estimates how much of the drive’s rated NAND endurance has been consumed and is the NVMe equivalent of wear leveling count on SATA SSDs. Watch available spare capacity declining toward its threshold, which signals NAND degradation. Track composite temperature and time spent above warning or critical temperature thresholds. Monitor media and data integrity errors, error log entries, and unsafe shutdown counts. Finally, alert on the critical warning bitfield, which covers available spare, temperature, reliability, read-only mode, and volatile memory backup failure in a single field.

Can I monitor NVMe SSD health from Linux?

What is the difference between SMART on SATA SSDs and NVMe drives?

NVMe uses a different health information log, not ATA-style SMART attributes. Instead of reallocated sector counts and vendor-specific fields, NVMe exposes percentage used, available spare, media errors, and a critical warning bitfield. smartmontools notes that NVMe drives do not provide ATA-style vendor attributes at all. The practical consequence: tools that parse ATA attributes may show incomplete or empty data for NVMe drives, so check that your tool reads the NVMe log natively.

How often should I check NVMe SSD health?

Match the cadence to the failure mode. Real-time monitoring at per-second to 60-second intervals catches thermal spikes and media error bursts as they happen. Daily or weekly audits are enough for slow-moving endurance and spare-capacity trends. For reference: Scrutiny defaults to daily collection, the Zabbix SMART template refreshes every 6 hours, smartctl_exporter polls every 60 seconds, and the Netdata NVMe collector defaults to 10 seconds. smartd can also schedule short and long self-tests on a regular cadence as a complementary layer.

Can Netdata monitor NVMe SSD health?

Yes. Netdata has a dedicated NVMe collector that reads estimated endurance, available spare, composite temperature, IO transferred, power cycles, critical warnings, unsafe shutdowns, media errors, and thermal management transitions. A separate SMART collector covers non-NVMe drives via smartctl. The NVMe collector defaults to a 10-second interval and is tunable, and a built-in alert fires on the NVMe critical warnings state. It requires nvme-cli and privileged access (ndsudo) on each host.

What is the best free NVMe SSD monitoring tool?

Netdata’s open-source agent and smartmontools are the best free NVMe SSD monitoring tools. Netdata offers a dedicated NVMe collector plus a free Cloud tier for small fleets; smartmontools is the lightweight CLI foundation that most other tools on this list wrap underneath. Other strong open-source options include Zabbix, Prometheus with smartctl_exporter, and Scrutiny. For Windows desktops, CrystalDiskInfo is the best freeware choice. Note that free tiers of commercial tools are capped by host, sensor, or service count, so ‘free’ rarely survives fleet growth.

How do I get alerts when an NVMe SSD is about to fail?

Set thresholds on percentage used, temperature, media errors, and the critical warnings bitfield. How much work that takes depends on the tool: the Zabbix SMART template ships triggers for endurance over 90%, high temperature, and smartctl exit-status bits; Netdata has a built-in alert on the NVMe critical warnings state; smartd can email alerts on health and threshold breaches. With smartctl_exporter and Prometheus, you write your own PromQL alerting rules.

Do I need a separate tool for NVMe vs SATA SSD monitoring?

Usually no, but the attribute sets differ and tool handling varies. Most tools cover both via smartctl. Netdata uses a dedicated NVMe collector plus a separate SMART collector for other drives. The Zabbix SMART template discovers both NVMe and SATA disks but disables self-test items for NVMe. Windows utilities like CrystalDiskInfo support both but present different attribute views. The key check is whether your tool reads the NVMe health log natively rather than reusing an ATA parser.

What does percentage used mean in NVMe SMART data?

It estimates how much of the drive’s rated NAND endurance has been consumed, making it the NVMe counterpart to wear leveling count on SATA SSDs. Reaching 100% means the drive has consumed its rated endurance, not that it will fail immediately - drives often keep working past it, but without a warranty-backed endurance guarantee. It is a planning signal: Zabbix triggers a warning when percentage used exceeds 90%, and Netdata reports the same data as estimated endurance percentage remaining.

The best NVMe and SSD health monitoring tools, ranked

How we evaluated NVMe and SSD monitoring tools

Netdata

Zabbix

Prometheus + smartctl_exporter

Checkmk

smartmontools

Scrutiny

PRTG Network Monitor

Nagios + check_smart

CrystalDiskInfo

Frequently asked questions

Book Your Free Demo