Buyer’s Guide - August 2026

The 10 best IoT device monitoring tools, ranked

Q: What is IoT device monitoring and how is it different from device management?

Monitoring watches the health, performance, and availability of devices: metrics, alerts, and dashboards. Device management handles provisioning, OTA firmware updates, and device identity. Most teams need both. This guide covers the monitoring half and notes where a management platform is required alongside it.

Q: Can I monitor IoT devices that publish over MQTT?

Yes, but the path differs by tool. Zabbix, PRTG, and ThingsBoard ingest MQTT natively. Prometheus needs an MQTT exporter, and Netdata needs a gateway application or exporter feeding the agent. Checkmk and PRTG also offer MQTT broker health checks, which watch the broker rather than the device topics themselves.

Q: What is the best way to monitor Raspberry Pi and ARM edge devices?

Agent footprint decides it. Netdata’s minimal mode runs under 1% CPU and 100MB RAM with full ARM support. Datadog’s IoT Agent supports armv7 and armv8 Linux. Zabbix Agent 2 officially supports Raspberry Pi. Uptime Kuma runs on a Pi 3 at roughly 80MB RAM for availability checks. All four are documented paths; pick based on whether you need metrics, protocols, or just uptime.

Q: Is there a free IoT monitoring tool?

Yes, several free IoT monitoring tools are available. Netdata has a free Cloud tier up to 5 nodes plus a free open-source agent. Zabbix, Prometheus, Nagios Core, Uptime Kuma, and LibreNMS are open source and self-hosted. Free means no license fee, not no operating cost: you provide the hardware, the database, and the administration time.

Q: What metrics should I monitor on IoT devices?

Start with CPU, memory, disk, and network on gateways and edge computers. Add temperature, battery, and power via sensor interfaces such as 1-Wire, lm-sensors, and IPMI. Track connectivity and availability: uptime, packet loss, and offline periods. Finally, watch for sensor drift and anomalies, which static thresholds routinely miss and ML-based detection catches.

Q: Can Netdata monitor thousands of IoT devices?

Yes. Each device runs the lightweight agent collecting per-second, and the parent-child architecture aggregates hundreds of devices per parent while buffering locally through outages. The design is proven on deployments exceeding 100K devices, and per-node pricing includes unlimited metrics, so fleet growth does not multiply the telemetry bill.

Q: Do I need a separate tool for OTA updates and device provisioning?

Yes. Monitoring tools, including Netdata, do not provision devices or push firmware. Pair your monitoring stack with a device management platform such as AWS IoT Device Management, Azure IoT Hub, or Balena. Netdata’s own IoT monitoring page explicitly recommends this pairing rather than pretending one tool covers both jobs.

Monitoring a fleet of sensors, gateways, and ARM edge computers is not the same job as monitoring servers. Devices are constrained, scattered across flaky field networks, and often speak MQTT, Modbus, or SNMP instead of running standard agents. This ranking grades ten tools on edge footprint, IoT protocol coverage, data granularity, fleet architecture, cost predictability, and time to value.

Start with Netdata (free) Jump to rankings

Why this list exists

IoT device monitoring is a different discipline from server monitoring. The targets are resource-constrained ARM boards and gateways, they sit behind NAT and cellular links that drop for hours, and half of them do not run a full OS at all. A tool that works beautifully in a datacenter can be useless, or ruinously expensive, across five thousand field devices.

The most expensive mistake buyers make here is conflating two categories. Device management platforms handle provisioning, OTA firmware updates, and device identity. Monitoring tools watch health, performance, and availability. Most teams need both, and none of the tools on this list, including our own, replaces the management half. This guide ranks the monitoring half and flags where pairing is required.

Three dimensions decide most outcomes:

Edge footprint and architecture. Can the agent run on a Raspberry Pi-class device without starving the workload, and does the topology buffer data locally when the network drops?
Protocol coverage and granularity. Native MQTT, Modbus, and SNMP ingestion versus exporter workarounds, and per-second collection versus 60-second polling that misses transients.
Cost shape at fleet scale. Per-node pricing stays predictable. Per-metric, per-series, and per-sensor models grow with exactly the thing IoT fleets produce: volume.

On pricing: this guide does not quote list prices for any vendor except Netdata. List prices change, volume discounts are negotiated, and the number that matters is what your fleet size and metric volume do to the bill. Each card links the vendor’s official pricing page so you can model your own deployment.

The shortlist was assembled from practitioner roundups (Logit.io, Censinet, Comparitech), vendor documentation, and community threads where engineers report what they actually run on Pi fleets and industrial gateways. We excluded pure device-management platforms that do not monitor, and included one IoT platform (ThingsBoard) because buyers consistently shortlist it alongside monitoring tools and deserve a clear answer on where it fits.

Edge and constrained-device support carries the most weight (25%), because an agent that cannot run on the device forces a gateway architecture for everything. IoT protocol coverage (20%) and data granularity (15%) follow, since they determine what you can see and how fast. Scoring draws on official documentation and published deployment patterns, not vendor marketing claims.

Tester credit

Compiled by the Netdata team - Updated August 12, 2026

Sources verified

Scoring criteria

Edge and constrained-device support 25%
Agent footprint on ARM and Pi-class hardware
IoT protocol coverage 20%
Native MQTT, Modbus, SNMP versus exporters
Data granularity 15%
Per-second versus 15-60s polling
Fleet scalability and edge architecture 15%
Offline buffering, parent-child topology
Cost predictability 15%
Per-node versus per-metric and per-sensor models
Time to value 10%
Install effort and out-of-box dashboards

Vendor 01 / 10 · #netdata

Netdata

Open-source, per-second infrastructure monitoring with a lightweight agent built to run on ARM and constrained edge devices.

Best for

Teams monitoring Raspberry Pi, BeagleBone, and industrial gateways at the edge
Fleets of thousands of distributed devices that need per-second visibility without a metrics bill that scales with volume
Organizations with data-sovereignty requirements that want metrics processed on-device with zero egress

Pricing

Per-node pricing: Netdata Cloud Business starts at $4.50/node/month on annual plans, with the per-node price decreasing as node count grows
Agents are open source and free; free Cloud tier covers small fleets up to 5 nodes
Homelab plan at $90/year for unlimited nodes; Enterprise on-prem from 200 node licenses
Unlimited metrics, logs, users, and retention per node - no per-GB or per-series charges

Pros

Per-second collection with auto-discovery; roughly 60-second install and zero-config dashboards
Minimal-mode footprint under 1% CPU and 100MB RAM; ARM fully supported for Pi, BeagleBone, and industrial gateways
SNMP auto-discovery plus native collectors for 1-Wire (DS18B20), lm-sensors, IPMI, and UPS hardware
Edge-native parent-child architecture buffers locally and replays on reconnect, keeping fleets observable on flaky networks
18-model ML anomaly detection with 99% false-positive reduction and AI root-cause analysis
Proven on 100K+ device deployments

Where teams pair it

No native MQTT, CoAP, or LoRaWAN ingestion - those protocols need a gateway application or exporter feeding the agent
No device provisioning, OTA firmware updates, or device identity - pair with AWS IoT Device Management, Balena, or Azure IoT Hub
Agent requires a full OS (Linux, Windows Server 2016+, macOS, FreeBSD); no Android or RTOS/bare-metal support

Verdict

Netdata leads this list because it was engineered for the edge rather than adapted to it. The agent runs on ARM gateways and sensors with a sub-1% CPU, sub-100MB minimal-mode footprint, collects per-second so sensor drift and connectivity blips are visible instead of averaged away, and the parent-child topology keeps a distributed fleet observable through network outages. Per-node pricing with unlimited metrics means the bill grows with device count, not with how much telemetry each device produces. The honest boundary: MQTT and LoRaWAN payloads arrive via exporters or gateways, and provisioning and OTA belong to a device management platform you run alongside it.

Start free -> See IoT monitoring ->

Vendor 02 / 10 · #zabbix

Zabbix

Open-source enterprise monitoring with native MQTT and Modbus support and an agent that runs on Raspberry Pi.

Best for

Teams that want a self-hosted platform with no license fee and native IoT protocol support
Organizations standardizing on MQTT or Modbus for device telemetry
Large fleets where per-node SaaS pricing would be prohibitive

Pricing

Open source and self-hosted: you run and operate it, with no license fee
Optional paid support subscriptions with SLAs and long-term maintenance
Zabbix Cloud available as a managed SaaS alternative

Pros

Native MQTT and Modbus monitoring built into Agent 2, with LoRaWAN via gateways
Agent 2 officially supported on Raspberry Pi
Mature alerting, templating, and auto-discovery; scales to tens of thousands of devices
Fully open source under AGPLv3 since version 7.0

Cons

Polling-based architecture with default 1-minute intervals; per-second collection requires tuning
Steep learning curve: configuration, templating, and maintenance are manual
You own the server, database, and HA burden; no provisioning or OTA features

Verdict

Zabbix is the strongest open-source alternative for IoT monitoring. Native MQTT and Modbus ingestion in Agent 2, official Raspberry Pi support, and zero license fees make it the default answer for teams that want protocol coverage without a SaaS bill. The trade-offs are operational: you run the server and database yourself, the configuration model rewards experienced admins and punishes newcomers, and polling granularity defaults to a minute, so transient sensor anomalies need deliberate tuning to catch. For protocol-heavy fleets with in-house operations skill, it earns rank two.

Read full Netdata vs Zabbix -> Zabbix pricing ->

Vendor 03 / 10 · #datadog

Datadog

Commercial observability platform with a dedicated IoT Agent and a per-device IoT plan for distributed fleets.

Best for

Enterprises already on Datadog that want device telemetry beside their cloud stack
Fleets of Linux ARM devices (armv7/armv8) that can run the optimized IoT Agent
Teams that want fully managed SaaS with no self-hosted infrastructure

Pricing

IoT plan billed per device, separate from per-host infrastructure pricing
Usage-based modules (logs, metrics, traces) add cost as volume grows
The bill scales with both fleet size and telemetry volume

Pros

Dedicated IoT Agent optimized for constrained environments, packaged for x64, arm64, and ARMv7
Per-device IoT plan with lower ingest and retention allotments than standard hosts
Deep integrations for AWS IoT, Azure IoT Edge, and Cisco Meraki device logs
Mature alerting, dashboards, and anomaly detection on one platform

Cons

SaaS-only: telemetry leaves the edge, which conflicts with data-sovereignty or air-gapped requirements
Per-device plus per-module pricing grows with fleet size and data volume
IoT Agent is Linux-only (DEB/RPM); no Windows IoT or RTOS support
Default 15-second collection misses sub-minute transients that per-second tools catch

Verdict

Datadog is a credible IoT option thanks to the ARM-capable IoT Agent and a per-device plan that acknowledges devices are not servers. If your organization already lives in Datadog, adding device telemetry to the same dashboards and alert pipelines is genuinely convenient. The constraints are structural: it is SaaS-only, so every sample crosses the network to Datadog’s cloud; the agent is Linux-only at the edge; and costs compound across per-device and per-module dimensions as fleets grow. For cloud-centric enterprises that trade sovereignty for convenience, it earns its place.

Read full Netdata vs Datadog -> Datadog pricing ->

Vendor 04 / 10 · #prometheus-grafana

Prometheus + Grafana

Open-source metrics stack widely assembled by hand for edge and IoT monitoring.

Best for

Teams with in-house SRE skills who want full control of the monitoring pipeline
Edge deployments that already export Prometheus-format metrics
Organizations standardizing on Grafana for visualization across stacks

Pricing

Both components are open source and self-hosted: you run and operate them
Grafana Cloud SaaS billed per active series and per GB of logs and metrics
Hosted cost grows with series cardinality and retention

Pros

Huge exporter ecosystem, including MQTT and Modbus exporters for device telemetry
Proven edge pattern: node_exporter and custom exporters on gateways, scraped over the network (documented by balena and MQTT broker vendors)
Grafana is the de-facto dashboard layer with thousands of community dashboards
Fully open source (Prometheus Apache-2.0, Grafana OSS AGPLv3)

Cons

DIY stack: you assemble scraping, retention, alerting, and HA yourself
Pull-based scraping is awkward across NAT’d or flaky field networks; push requires extra components
No native MQTT ingestion in Prometheus - needs an exporter or gateway
Cardinality and retention management are manual; hosted Grafana bills per active series

Verdict

Prometheus plus Grafana is the default DIY choice for edge monitoring, and for good reason: the exporter ecosystem covers nearly everything, and the operational patterns for gateway scraping are well documented. It monitors IoT well when your devices already speak the format and your network cooperates. Where it struggles is exactly where IoT fleets live: pull-based scraping across NAT and intermittent links, no native MQTT, and cardinality discipline that becomes a full-time concern at fleet scale. Budget engineering time, not just infrastructure.

Read full Netdata vs Prometheus -> Grafana pricing ->

Vendor 05 / 10 · #checkmk

Checkmk

Open-core monitoring platform with a dedicated IoT monitoring solution and MQTT broker checks.

Best for

IT teams monitoring IoT devices alongside servers and network gear in one platform
Environments using MQTT brokers whose health and message flow need watching
Buyers who want commercial support on top of an open-source core

Pricing

Open core: Raw edition is open source and self-hosted
Commercial subscriptions (Cloud, Pro, Ultimate) priced per monitored host
Cost grows with host count

Pros

Dedicated IoT monitoring solution covering building automation, environmental sensors, smart meters, and lighting
MQTT broker checks (clients, messages, uptime) via the agent_mqtt special agent
Agent supports ARM platforms for edge devices
Auto-discovery and a polished web UI reduce configuration effort

Cons

MQTT support focuses on broker health rather than ingesting arbitrary device topics
Default 1-minute check intervals; per-second monitoring is not the model
Full-featured editions are per-host subscriptions; Raw edition lacks enterprise features
Device provisioning and OTA are out of scope

Verdict

Checkmk treats IoT as one workload in a broader estate, and does it well: the dedicated IoT solution and MQTT broker checks give real visibility into message infrastructure, and auto-discovery keeps configuration sane. Its limitation is depth on the device side. Broker health is not the same as ingesting arbitrary device telemetry, and the one-minute check model is coarse for sensor behavior. A solid enterprise pick when IoT shares a console with servers and switches.

Read full Netdata vs Checkmk -> Checkmk pricing ->

Vendor 06 / 10 · #prtg

PRTG (Paessler)

Windows-based all-in-one network monitor with 250+ sensor types, including dedicated IoT and MQTT sensors.

Best for

Windows-centric IT teams wanting one tool for network, server, and IoT monitoring
Facilities and industrial sites using Sigfox or MQTT push telemetry
Small-to-mid fleets where the sensor-based model stays predictable

Pricing

Licensed per sensor count in tiers, not per device
Freeware edition available with a sensor cap
Perpetual license or annual subscription; cost grows with sensor count

Pros

Dedicated IoT sensors: MQTT Round Trip, HTTP IoT Push Data Advanced (Sigfox-ready), and MQTT/Modbus options
250+ native sensor types with auto-discovery
Remote probes extend monitoring to distributed sites without per-device agents
Proven in IoT case studies such as environmental monitoring at the National Museum of Computing

Cons

Server runs on Windows only, ruling out Linux-only shops
Per-sensor licensing penalizes devices that expose many metrics
Default 60-second scanning; sub-minute monitoring needs tuning
No native MQTT topic ingestion beyond broker and push sensors

Verdict

PRTG is strong where IoT meets the network closet: SNMP devices, Sigfox push endpoints, and MQTT broker health, all from a single Windows console. The sensor model is easy to reason about until it is not - a gateway exposing forty metrics consumes forty sensors, and fleets with rich telemetry hit license tiers fast. Combined with Windows-only deployment and minute-scale defaults, it fits facilities and mid-size sites better than large distributed edge fleets.

Read full Netdata vs PRTG -> PRTG pricing ->

Vendor 07 / 10 · #nagios

Nagios

The classic open-source monitoring engine, widely run on Raspberry Pi to watch SNMP and agent-based devices.

Best for

Teams that want a battle-tested, plugin-driven monitor with a huge community
Raspberry Pi-based monitoring of SNMP-capable devices
Organizations already invested in Nagios plugins and NRPE

Pricing

Nagios Core is open source and self-hosted: you run and operate it
Nagios XI is commercial, licensed per monitored node
Cost grows with node count on XI

Pros

Runs on Raspberry Pi and monitors other Pis and SNMP devices via check_snmp and NRPE
Nagios Exchange has thousands of plugins, including Modbus TCP/RTU and Pi temperature checks
Core is open source (GPL) with no license fee
Decades of documentation and community knowledge

Cons

Configuration is file-based and manual; the UI and dashboards feel dated
Polling-based with default 60-second checks; no per-second granularity
Nagios XI per-node licensing adds cost at fleet scale
No native MQTT or Modbus in the core - relies on community plugins

Verdict

Nagios remains a viable low-cost option for small fleets of SNMP-capable devices, and its Pi-friendliness is genuinely proven. The plugin archive covers almost anything with a check script. But the architecture shows its age: file-based configuration, minute-scale polling, and IoT protocol support that depends entirely on community plugins of varying quality. Choose it when you already speak Nagios; starting fresh in 2026, the newer tools on this list will cost you less time.

Read full Netdata vs Nagios -> Nagios pricing ->

Vendor 08 / 10 · #thingsboard

ThingsBoard

Open-source IoT platform for device connectivity, telemetry collection, alarms, and dashboards.

Best for

Product teams building IoT solutions that need connectivity plus monitoring in one platform
Deployments speaking MQTT, CoAP, or HTTP that want rule chains and alarms
Teams wanting an open-source platform with an active LTS release cadence

Pricing

Community Edition is open source and self-hosted (Apache-2.0)
Professional Edition is a per-instance commercial subscription
ThingsBoard Cloud is a hosted SaaS option; cost scales with device count

Pros

Native MQTT, CoAP, and HTTP device connectivity with telemetry storage
Built-in alarms, rule chains, and customizable dashboards for device health
Gateway supports Modbus, BACnet, OPC-UA, BLE, and CAN bus for legacy devices
Actively maintained with LTS releases in the v4.x line

Cons

It is an IoT platform, not a monitoring tool: you build the monitoring experience from telemetry and rules
No agent for constrained devices - devices must publish to it
Self-hosted operation is heavy: Java stack, database, HA planning
PE features such as white-labeling and advanced RBAC require a commercial license

Verdict

ThingsBoard is the right pick when you are building an IoT product and want connectivity, storage, rule engines, and dashboards as one platform. Its native protocol support is the broadest on this list. But it inverts the monitoring model: nothing watches the device itself unless the device publishes to ThingsBoard, and there is no agent seeing CPU, memory, or thermals on the hardware. Many teams end up running a monitoring tool alongside it, which is a legitimate architecture - just know that going in.

Visit ThingsBoard -> ThingsBoard pricing ->

Vendor 09 / 10 · #uptime-kuma

Uptime Kuma

Lightweight self-hosted uptime monitor that runs happily on a Raspberry Pi and watches device endpoints.

Best for

Hobbyists and small teams monitoring a handful of IoT endpoints from a Pi
Status-page-style visibility of device APIs, MQTT brokers, and web endpoints
Anyone who wants push-based monitoring with many notification channels

Pricing

Open source and self-hosted: no license fee, no paid tiers
Cost is the hardware and time you run it on

Pros

Single lightweight container idling around 80MB RAM - runs on a Pi 3
Monitors HTTP, TCP ports, ping, DNS, and push heartbeats with a clean dashboard
90+ notification integrations including Telegram, Discord, and email
MIT-licensed open source with an active community

Cons

Availability monitoring only: no CPU, memory, temperature, or sensor metrics
No SNMP, MQTT, or Modbus support - devices must expose HTTP, TCP, or ping
Single-node design; no fleet topology, RBAC, or multi-tenant features
Not designed for thousands of devices or long-term metric history

Verdict

Uptime Kuma is delightful at what it does: telling you whether a device endpoint is alive, from hardware that costs less than dinner. For small IoT setups and home labs, it is often the right first tool. Its limits are definitional - it cannot see inside a device, speaks no IoT protocols, and has no fleet story. Treat it as an availability layer, and pair it with a metrics tool when the fleet grows past a shelf of devices.

Read full Netdata vs Uptime Kuma -> Uptime Kuma on GitHub ->

Vendor 10 / 10 · #librenms

LibreNMS

Open-source network monitoring with automatic SNMP discovery, popular for watching network-connected IoT devices.

Best for

Network teams monitoring SNMP-capable IoT devices: switches, PDUs, UPS units, sensors
Organizations that want auto-discovery without per-device agents
Budget-conscious shops preferring a community-supported GPL platform

Pricing

Open source and self-hosted: you run and operate it, with no license fee
Cost is infrastructure and administration time

Pros

Automatic SNMP discovery (v1/v2c/v3) with zero MIB hunting for common devices
Community fork of Observium with active development and a large install base
Alerting, graphing, and API built in; runs on modest hardware
GPLv3 open source with no license fees

Cons

SNMP-only focus: no MQTT, Modbus, or agent-based collection for non-SNMP devices
Default 5-minute polling is coarse for transient device failures
Requires SNMP on the device - many constrained IoT devices do not expose it
No device management, OTA, or fleet provisioning features

Verdict

LibreNMS is excellent inside its lane: point it at a network and it discovers every SNMP-speaking device, maps topology, and starts graphing with almost no configuration. For gateways, PDUs, UPS units, and industrial sensors that expose SNMP, it is a proven free-to-license choice. Outside that lane it simply does not operate - no agents, no MQTT, and five-minute polling that smooths away the transients IoT fleets produce. Rank it highly for network-adjacent IoT; look elsewhere for the devices themselves.

Read full Netdata vs LibreNMS -> LibreNMS site ->

Frequently asked questions

What is IoT device monitoring and how is it different from device management?

Can I monitor IoT devices that publish over MQTT?

What is the best way to monitor Raspberry Pi and ARM edge devices?

How much does IoT device monitoring cost?

IoT device monitoring ranges from free open-source tools to per-device commercial plans, and the pricing model matters more than the sticker price. Models differ: per-node (Netdata), per-device (Datadog IoT), per-sensor (PRTG), per-host (Checkmk), and per-active-series (Grafana Cloud). Open-source tools such as Zabbix, Prometheus, Nagios Core, Uptime Kuma, and LibreNMS carry no license fee, but you run and operate them yourself. Whatever model you choose, model the bill against your fleet size, metric volume, or sensor count before committing.

Is there a free IoT monitoring tool?

What metrics should I monitor on IoT devices?

Can Netdata monitor thousands of IoT devices?

Do I need a separate tool for OTA updates and device provisioning?

The 10 best IoT device monitoring tools, ranked

How we evaluated IoT monitoring tools

Netdata

Zabbix

Datadog

Prometheus + Grafana

Checkmk

PRTG (Paessler)

Nagios

ThingsBoard

Uptime Kuma

LibreNMS

Frequently asked questions

Book Your Free Demo