For a team with somebody to own it, Zabbix is the strongest thing here: no license fee at any size, support priced on response time rather than device count, and a published five year lifecycle almost no commercial vendor matches. For depth on why the network is slow, SolarWinds Observability Self-Hosted.
For a managed service provider, Auvik, because configuration backup is included and that is the feature that ends outages, though it publishes no price at all. If you want a number today, PRTG is the only one whose whole list is on the page.
- Best overall, and free: Zabbix
- Most depth on why it is slow: SolarWinds Observability Self-Hosted
- Best for MSPs, if a sales call is acceptable: Auvik
- Clearest pricing, in the trickiest unit: PRTG
- Cheapest route to configuration backup: Domotz
We do not accept payment for placement. Scores come from a published 40-point rubric, applied to vendor pricing pages, product documentation and license files read at the source on September 10, 2026; vendors can dispute a fact but not a score. How we rank software · Affiliate disclosure
The rankingComparison table
| Tool | Best for | Pricing model | Starts at | Score |
|---|---|---|---|---|
 ZabbixEditor’s pick | Teams with somebody to own it | Free, support sold separately | $0 | 8.9 |
 SolarWinds Observability Self-HostedMost depth | Large networks that need path level answers | Per node | $8 per node / mo | 8.4 |
 LogicMonitor | Hybrid estates where the network is half the problem | Per hybrid unit | $16 per unit / mo | 8.0 |
 AuvikBest for MSPs | MSPs managing networks they did not build | Quote only | Not published | 7.6 |
 ManageEngine OpManager | Small networks that want a commercial product | Per device | $245 for 25 devices | 7.5 |
 LibreNMS | Network teams who want discovery to just work | Free, GPL-3.0 | $0 | 7.4 |
 PRTG Network MonitorClearest pricing | Small networks that want to know the price today | Per sensor | $200 / mo for 500 sensors | 7.3 |
 Datadog Network Device Monitoring | Teams already running Datadog | Per device | $7 per device / mo | 7.1 |
 DomotzBest value at one site | Many small sites, one flat price each | Per collector or per device | $35 per collector / mo | 6.5 |
 Progress WhatsUp Gold | Windows shops that want to own the license | Per device, banded | $1,229 for 50 devices / yr | 6.4 |
Score is out of 10, from the 40 point rubric below. Every price was read at the vendor’s own pricing page on September 10, 2026, and every license at the project’s own repository, so any number here can be checked against the page it came from.
Network monitoring tools, or network management software?
Both phrases return this same list of vendors. Network monitoring tools is the narrower of the two: watch the equipment, say when something is down or slow. Management adds the third job, changing and restoring configuration. Every product here monitors. Six of them stop there in what you first buy.
So a search for the best network monitoring software and a search for the best network management software land on the same ten products, and the difference only shows up in the quote. That is why this page scores the three jobs separately rather than ranking on a single word.
If monitoring is all you need, the cheap end of the table below is genuinely cheap and the open source options are genuinely complete. If somebody will one day ask what changed on that switch, you are shopping for management and the shortlist gets much shorter.
The unit is the decision
There is a problem with this category that no comparison table normally admits. Every vendor prices in a different unit. PRTG sells sensors. SolarWinds sells nodes. OpManager and Datadog sell devices. Domotz sells collectors, or devices. LogicMonitor sells a hybrid unit it does not define. Zabbix sells response time and gives the software away. Auvik sells nothing you can see.
So a headline price comparison across this list is arithmetic performed on incompatible things. The only way to compare them is to take one network and price it in every vendor’s own unit. Here is a 200 device network, using each vendor’s published figures and each vendor’s own conversion where they give one.
| Tool | What it sells | 200 devices, at list | What is not in that number |
|---|---|---|---|
| Zabbix | Response time | $0, or $3,900 a year with Silver support | The person who runs it, and Oxidized for config |
| LibreNMS | Nothing | $0 | The person who runs it. No support is sold |
| Domotz | Collectors or devices | $3,600 a year at $1.50 a device | Little. Config backup is included |
| PRTG | Sensors | $8,904 a year, at Paessler’s 5 to 10 sensors a device | Config backup, which the product does not do |
| Datadog | Devices | $16,800 a year at $7 a device | The infrastructure hosts that do the polling |
| SolarWinds | Nodes | $19,200 a year at $8 a node, before discount | NetFlow Traffic Analyzer and Network Configuration Manager |
| OpManager | Devices, banded | from $4,595 at the 250 device band | NetFlow Analyzer from $172, and Network Configuration Manager |
| WhatsUp Gold | Devices, banded | Not published above 50 devices | The configuration management add-on |
| LogicMonitor | A hybrid unit | Cannot be calculated, the unit is undefined | Unknown until the unit is defined |
| Auvik | Not disclosed | Not published | Everything |
The same 200 devices, from nothing to nineteen thousand dollars a year, and three of the ten cannot be priced from public information at all. Two things follow. The first is that any shortlist built from headline prices is built on sand.
The second is subtler and it is the reason PRTG sits where it does: the sensor is the only unit on that list that changes with how carefully you monitor, so the more thorough you are, the more you pay. Every other unit is fixed by your inventory.
What each one actually answers
The three questions this category exists to answer are not the same product. Is it up comes from polling. Why is it slow needs flow data and path measurement.
What changed needs a stored copy of every configuration and something that compares them. The third is the cheapest to run and the one that ends outages fastest, and this is where it lives in each product.
| Tool | Is it up | Why is it slow | What changed |
|---|---|---|---|
| Zabbix | Included | Partial, not a first class feature | Not native. Oxidized alongside |
| SolarWinds | Included | Included, and the deepest here | A separate product, Network Configuration Manager |
| LogicMonitor | Included | Included | Included in the platform |
| Auvik | Included | Included | Included, and the reason to buy it |
| OpManager | Included | A separate product, NetFlow Analyzer | A separate product, Network Configuration Manager |
| LibreNMS | Included | Partial | An integration, Oxidized, documented |
| PRTG | Included | Included, through flow sensors | Not in the product |
| Datadog | Included | Included, with stated retention | Not offered |
| Domotz | Included | Partial | Included, with restore |
| WhatsUp Gold | Included | Included | An add-on module |
Six of the ten do not answer the third question in what you first buy. That is the single most useful line on this page. Nothing breaks while configuration backup is switched off, which is exactly why it is the thing left out of the quote and the thing nobody notices is missing until the afternoon somebody needs it.
MethodHow we picked
Forty points across six criteria, the same six the rest of this site uses, with each one read against what this category is for. Core depth is question one: discovery, protocol coverage, and what the poller can be made to watch.
Secondary depth is question two: flow records, path measurement, and whether the tool can tell you why rather than whether. Automation is question three plus what the product will do without a human.
Pricing clarity is whether a buyer can find out the price. Scale is what the vendor documents rather than what it implies. Support and longevity is what you can buy and what the vendor commits to in writing.
Where a figure came from is on the page next to it. Prices are the vendor’s own published list, read at their pricing page and dated. Licenses are the SPDX identifier in the project’s own repository, not the one the internet repeats, which is how Icinga 2 turns out to be GPL-3.0 rather than the GPL-2.0 every roundup copies from the last one.
Where a vendor publishes nothing, the page says so and the pricing clarity score reflects it, because refusing to publish a price is a fact about the product.
The thing no ranking scores
A network management server holds credentials to every device it touches. It is, on most networks, the single most privileged machine in the building, and its own supply chain is therefore part of your attack surface.
CISA put the point plainly in Emergency Directive 21-01, which ordered federal agencies to power down SolarWinds Orion versions 2019.4 through 2020.2.1 HF1 in 2020: access to that server permits an attacker to gain access to network traffic management systems. The directive is closed and the activity was later attributed to the Russian SVR.
The reason to raise it here is not that one vendor is uniquely risky. It is that the risk belongs to the category.
Whichever product on this page you choose, the questions worth asking are where its credentials are stored, what the update channel is, whether the server needs outbound internet access, and who can log in to it. None of those are on a feature matrix and all of them matter more than the dashboard.
What the monitoring half actually does
Under the marketing, network monitoring tools do four mechanical things. They poll equipment over SNMP for interface counters, CPU, memory and temperature. They test reachability with ICMP. They receive what a device sends unasked, SNMP traps and syslog. Then they compare all of it against a threshold and raise an alert.
The vendors describe it in those terms themselves. Zabbix documents SNMP v1, v2c and v3 with trap collection, polling or traps, device availability and uptime, and flexible problem thresholds. LogicMonitor states SNMP, WMI or API polling for CPU, bandwidth and latency. Domotz monitors any SNMP OID and sets thresholds on it.
Bandwidth is the part people mean by performance monitoring, and it runs on a different mechanism: flow records exported by the device, NetFlow or sFlow, rather than a counter read on a schedule. That is why question two costs extra in so many of these products, and why the tables above keep it separate.
Polling interval is the setting that decides what the data is worth. A five minute poll cannot see a thirty second outage and no dashboard recovers it afterwards. Ask what interval a vendor assumes in their sizing at your device count, before comparing two graphs drawn at different rates.
What question two is measuring
Is it up is a yes or a no. Why is it slow is a set of numbers: response time, latency, packet loss, and which conversation is filling the link. Progress states that WhatsUp Gold alerts and reports on performance and response times and identifies bandwidth hogs. LogicMonitor names latency and top talkers.
Those numbers are only worth anything against a baseline, which is the real argument for retention. A graph reaching back a week says today is bad. A graph reaching back a year says when it started, and that is usually the question in the room. Read the retention period in the quote.
Alerting is where these deployments fail, and the vendors know it. ManageEngine sells threshold based performance monitors, Progress markets topology aware alerting that suppresses the storm behind one failed uplink, and LogicMonitor sells dynamic thresholds for the same reason. Progress also lists delivery by SMS, email, web or Slack.
Cloud based, or a server you run
The other split in this list is where the software lives, and for a small team it decides more than the feature grid does. Auvik describes itself as a cloud based network management solution: a free collector sits on site and proxies to the cloud, discovering devices with ping and SNMP.
Domotz works the same way and publishes its deployment options, from a Domotz Box to a VM, Windows, Linux or a NAS. Zabbix, LibreNMS, OpManager and WhatsUp Gold are servers you stand up, though Zabbix also publishes Zabbix Cloud and Paessler offers PRTG Hosted Monitor beside the on-premises build.
The trade is the familiar one. Cloud based means no server to patch and a collector that dials out. Self-hosted means the monitoring data stays on your own infrastructure and the upgrade is your problem. Ask a cloud vendor what alerting does when the site link drops, because the collector is inside the outage.
What the scores are made of
40 points across 6 criteria, applied to vendor pricing pages, product documentation and license files read at the source on September 10, 2026.
Zabbix
The argument for putting a free product at the top of a paid category is not the price. It is that Zabbix publishes something almost nobody else here does: a lifecycle table with dates in it. An LTS every eighteen months, supported for five years, three of full support and two of limited.
Zabbix 7.0 LTS was released on June 4, 2024, full support runs to June 30, 2027 and limited support to June 30, 2029. 8.0 LTS is planned for Q3 2026. You can plan an upgrade cycle against that. Ask a commercial vendor on this page for the same table.
The commercial model is the other reason. Zabbix sells support, not licenses, and prices it on how fast somebody answers rather than on how many devices you monitor. The page says so in its own words: no license fees, no growth tax, with unlimited hosts, devices, users, metrics and alerts on every tier.
Silver is EUR 245 or USD 325 a month billed annually for 8x5 with a one day response. Gold starts at EUR 660 or USD 825 for a four hour response. Doubling the size of your network changes nothing on that invoice, which is the opposite of every per-device product below.
What it costs instead is a person. Somebody installs it, upgrades it, writes the templates and tunes the thresholds, and that person is the real price. The honest test is not whether you can afford the license, because you can. It is whether that person already exists.
Where it wins
- A published five-year lifecycle with real dates, which most of the commercial field does not offer
- Support priced on response time rather than device count, so growth does not reprice you
- AGPL-3.0, verified in the repository COPYING file, with no license fee at any scale
- Proxies for distributed polling and native high availability
- A template library that covers most vendors without writing anything
Where it loses
- No configuration backup, so question three needs Oxidized running alongside it
- Flow analysis is not a first-class feature the way it is in the commercial tools
- The install, the upgrades and the threshold tuning are your labor, and that is the actual cost
- Support tiers above Gold are quote only
How Zabbix scores, criterion by criterion
Zabbix pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| The software | $0 | $0 | AGPL-3.0, unlimited devices, no license tier |
| Silver support | $325 a month | $3,900 | 8x5, one day response, unlimited hosts |
| Gold support | from $825 a month | from $9,900 | 8x5, four hour response, adds distributed monitoring |
| Platinum, Enterprise | Custom | Custom | 24x7, two hour and 1.5 hour response |
SolarWinds Observability Self-Hosted
First, a naming problem that costs buyers time. Network Performance Monitor still exists as a product page, and its own title now reads Network Performance Monitor - Observability Self-Hosted.
NPM is the network module of a larger platform, and the price on the SolarWinds pricing page is for that platform: from $8 per node per month, with volume discounts, bought through a quote.
What you get for it is the deepest answer to question two on this page. NetPath draws the route hop by hop across your own network and the internet beyond it, which is the difference between knowing a link is busy and knowing which hop is losing packets.
PerfStack puts network, system and application metrics on one timeline. Nothing else in this ranking goes that far into the path.
The catch is structural. The three questions are three purchases: NPM for question one and part of two, NetFlow Traffic Analyzer for the traffic detail, Network Configuration Manager for question three. Each is a separate product with a separate page and a separate line on the quote, so the number you end up paying has very little to do with $8.
One more thing a buyer should weigh with open eyes, and it applies to every tool on this page rather than only this one. A monitoring server holds credentials to every device it touches, which makes it the most privileged box on the network.
The Orion platform this product line descends from was the vehicle of the 2020 supply chain compromise: CISA Emergency Directive 21-01 named versions 2019.4 through 2020.2.1 HF1, ordered federal agencies to power the servers down, and the activity was later attributed to the Russian SVR. The directive is closed and the product has moved on.
The reason to raise it is not that this vendor is uniquely risky, it is that the directive stated the general principle plainly: access to that server is access to network traffic management systems. That is true of Zabbix and LibreNMS too, and nobody scores it.
Where it wins
- NetPath hop by hop path visualization, which nothing else here matches
- PerfStack correlates network, system and application metrics on one timeline
- Per node licensing that can be reallocated across instances
- A 30 day fully functional trial, downloadable without a sales call
- Documented at sizes the rest of this list does not reach
Where it loses
- The three questions are three separate products, so the quote is never the headline price
- Price is published only as a starting figure, and buying goes through a quote
- Self-hosted, so the monitoring server and its database are yours to run and patch
- The module list is long enough that comparing it to a single price product is guesswork
How SolarWinds Observability Self-Hosted scores, criterion by criterion
SolarWinds Observability Self-Hosted pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Observability Self-Hosted | from $8 per node / mo | from $19,200 at 200 nodes | List, before volume discount. Bought via quote |
| NetFlow Traffic Analyzer | Separate product | Quote | Question two, in detail |
| Network Configuration Manager | Separate product | Quote | Question three |
| Trial | Free | Free | 30 days, fully functional |
LogicMonitor
LogicMonitor is the one commercial platform here that treats question three as part of the product rather than as another SKU.
Configuration Monitoring is a listed platform capability, described by the vendor as tracking device configuration drift to reduce change failure rates, and it sits next to topology and telemetry instead of behind a second purchase. On the argument this ranking is built on, that matters more than any dashboard in the category.
Three packages are published: Essentials at $16 per hybrid unit per month, Advanced at $27, and Signature with Edwin AI at $53. That is more disclosure than most of this list manages, and it stops one step short of being useful, because the pricing page does not define a hybrid unit.
The footnote adds that these are starting list prices based on standard minimum quantities and terms. So you know the shape of the bill and not the bill, which is why the pricing score here is 3 out of 5 rather than the 5 the published numbers would otherwise earn.
It fits an estate where the network is one of several things going wrong at once and the team wants one console for all of them. A network team that only owns switches and routers is buying a wider platform than it needs.
Where it wins
- Configuration drift tracking is in the platform, not sold as a module
- Topology aware correlation, which shortens the argument about what caused what
- Three package prices published, which most SaaS competitors do not do
- One console across cloud and on-premises, which suits a mixed estate
Where it loses
- The pricing page never defines a hybrid unit, so the published price cannot be turned into a quote
- Minimum quantities and contract terms apply, so list is a starting point
- SaaS only, with no self-hosted option
- The top package is more than three times the entry one
How LogicMonitor scores, criterion by criterion
LogicMonitor pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Essentials | $16 per hybrid unit / mo | Depends on the unit count | Starting list price, minimums apply |
| Advanced | $27 per hybrid unit / mo | Depends on the unit count | Enterprise grade tier |
| Signature with Edwin AI | $53 per hybrid unit / mo | Depends on the unit count | Adds the AIOps engine |
| Trial | Free | Free | 15 days, any plan |
Auvik
Auvik scores 7 out of 7 on automation and 1 out of 5 on pricing clarity, and those two numbers are the entire review.
It is the strongest answer on this page to the question that actually shortens outages. Configuration management is a first class feature, the topology map is built from discovery rather than from what somebody drew last year, and for a provider walking into a network they did not build, those two together are the product.
There is a real concession in the licensing too: access points, uninterruptible power supplies, printers and network attached storage are monitored without counting toward the license, which is unusual and worth money on a typical site.
Then there is the pricing page, which publishes no price. It is a form asking for first name, last name, business email, phone number, company name and employee count.
Its headline reads the only official source for Auvik pricing, and beneath that, competitors and AI tools show the wrong prices, get accurate pricing directly from Auvik.
Both sentences are true, and the reason competitors show the wrong prices is that the right ones are not published. You cannot put this product on a spreadsheet next to any other product on this page without a sales call first, and that is a decision the vendor made, not an accident.
Where it wins
- Configuration backup and change tracking included, which is the criterion that shortens outages
- Topology mapping built from live discovery rather than from documentation
- Access points, UPSs, printers and NAS monitored without counting toward the license
- A 14 day trial with no credit card and no feature paywall
- Packages built specifically for managed service providers
Where it loses
- No price is published anywhere, at any tier, in any currency
- The pricing page is a lead form that wants a phone number and an employee count
- Software as a service only, with no perpetual license
- Nothing about the cost can be compared to a competitor without talking to sales
How Auvik scores, criterion by criterion
Auvik pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Any tier | Not published | Not published | A form requiring name, email, phone, company, employee count |
| Perpetual license | Not offered | Not offered | Software as a service only, per the vendor |
| Free monitoring | Included | Included | APs, UPSs, printers and NAS do not count toward the license |
| Trial | Free | Free | 14 days, all features, no credit card |
ManageEngine OpManager
OpManager publishes its editions with device counts attached, which puts it ahead of most of this category before anything else is considered. Standard starts at $245 for 25 devices, Professional at $345 for the same 25, and Enterprise at $4,595 for 250. A small network can find out what this costs without speaking to anybody.
The licensing unit is the device and the vendor is pointed about it: a license covers the device inclusive of all its interfaces and sensors, and a device can have any number of interfaces, elements or sensors, unlike our competitors.
That is aimed squarely at sensor based pricing, and on a switch heavy network it is a fair shot. A 48 port switch is one device here and can be dozens of sensors elsewhere.
What pulls the score down is that the same modularity problem as SolarWinds applies at a smaller scale. Flow analysis is NetFlow Analyzer, a separate product starting at $172. Configuration backup is Network Configuration Manager, another separate product. The $245 answers question one, and questions two and three are two more purchases.
Where it wins
- Editions published with device counts, so a small buyer can self serve
- Per device licensing that includes every interface and sensor on that device
- Probe and central architecture for monitoring several sites from one console
- The lowest published entry price of any commercial product here
Where it loses
- Flow analysis is NetFlow Analyzer, a separate product from $172
- Configuration backup is Network Configuration Manager, a third product
- The step from Professional to Enterprise is $345 for 25 devices to $4,595 for 250
- Prices are all published as starting figures, so the quote is still the real number
How ManageEngine OpManager scores, criterion by criterion
ManageEngine OpManager pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Standard | from $245 | for 25 devices | Discovery, availability, interfaces, alerting |
| Professional | from $345 | for 25 devices | Adds network path analysis and workflow automation |
| Enterprise | from $4,595 | for 250 devices | Distributed monitoring across sites |
| NetFlow Analyzer | from $172 | Separate product | Question two is not in the box |
LibreNMS
If Zabbix asks you to describe your network, LibreNMS tries to work it out for itself. Auto discovery walks the neighbor protocols the devices already speak and builds the inventory from what it finds, which is why it is usually the open source tool that produces something useful on the first afternoon.
The release record is the reassuring part. Version 26.8.2 shipped on August 31, 2026, 26.8.1 on August 19 and 26.8.0 on August 17, which is a monthly train that has not slipped. The license is GPLv3 with exceptions for bundled third party packages, stated in the repository license file itself.
Question three has a documented answer: Oxidized, an Apache-2.0 project that retrieves and versions device configurations, with an integration page in the LibreNMS documentation rather than a forum post. That is two things to run instead of one, and it is a genuine answer rather than a gap.
What it does not have is a support commitment. There is no LTS, no published end of support date, and no tier you can buy to make somebody answer at three in the morning. For a team that is comfortable with that, it is the shortest path to a working monitoring system on this page.
Where it wins
- Auto discovery through the neighbor protocols, so the inventory builds itself
- A monthly release train with three releases in August 2026 alone
- Oxidized integration documented in the project docs, which answers question three
- GPL-3.0, no license cost, distributed pollers for larger networks
Where it loses
- No LTS and no published end of support dates, unlike Zabbix
- No commercial support tier from the project itself
- Flow analysis is not native, so question two stays shallow
- You own the upgrade path and whatever it breaks
How LibreNMS scores, criterion by criterion
LibreNMS pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| The software | $0 | $0 | GPL-3.0 with exceptions, per the repository license file |
| Oxidized for config backup | $0 | $0 | Apache-2.0, integration documented by LibreNMS |
| Support | None sold | None sold | Community only, which is the real cost |
| Your time | The actual price | The actual price | Install, upgrade, tune, and be the person who knows it |
PRTG Network Monitor
Paessler publishes the whole list, in dollars, with device equivalents next to each tier, and you can buy it from the page. In a category where the market leader says starts at and one competitor says nothing at all, that deserves the credit it gets here.
Then there is the unit. PRTG licenses sensors, where a sensor is one measured value: a switch port, a CPU, a disk. Paessler states the conversion on its own page: on average, you need about 5 to 10 sensors per device or one sensor per switch port.
That multiplier is the whole buying decision. PRTG 500 at $200 a month is described by the vendor as enough for about 50 devices, and if your network is switch heavy and you want per port visibility, a single 48 port switch can eat a tenth of that license.
Two more limits follow from the same design: the largest tier is 10,000 sensors for about 1,000 devices, and licenses cannot be combined on one server, so you upgrade rather than add.
The freeware edition deserves a mention because it is not a trial. One hundred sensors, about ten devices, free and permanent, which is a real monitoring system for a small office and the most generous no cost tier from any commercial vendor here.
What it will not do is answer question three. There is no configuration backup in the product, and the sensor list in the manual has nothing that covers it, so a network where the last outage came from a change somebody made is a network this tool cannot investigate.
Where it wins
- The complete price list published with device equivalents, purchasable from the page
- A permanent free tier of 100 sensors, which is a product rather than a trial
- One sensor per measured value gives genuinely fine grained control
- A 30 day unrestricted trial that falls back to freeware instead of stopping
Where it loses
- The unit is the sensor, and Paessler says a device is 5 to 10 of them
- A ceiling of 10,000 sensors per server, and licenses cannot be combined
- No configuration backup in the product at all
- Above about 1,000 devices you are moved to Enterprise Monitor, which is quote only
How PRTG Network Monitor scores, criterion by criterion
PRTG Network Monitor pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Freeware | $0 | $0 | 100 sensors, about 10 devices, permanent |
| PRTG 500 | $200 / mo paid annually | $2,400 | 500 sensors, about 50 devices |
| PRTG 2500 | $742 / mo paid annually | $8,904 | 2,500 sensors, about 250 devices |
| PRTG 10000 | $1,642 / mo paid annually | $19,704 | 10,000 sensors, about 1,000 devices |
Datadog Network Device Monitoring
Seven dollars a device a month is a clean number, published on a public page, in a unit a network engineer actually recognizes. Wireless access points are cheaper still at $4. Flow records are retained for 15, 30, 60 or 90 days with 30 as the default, which is more explicit than most competitors manage about the thing that drives storage cost.
The qualifier is in Datadog’s own FAQ and it changes the arithmetic: network device monitoring only requires the purchase of one or more infrastructure hosts which are used to poll devices.
The $7 buys the device, not the machine doing the polling, so the invoice has a second line on it that the headline does not mention. On demand pricing is another 45 percent above the annual rate, at $10.20 a device.
Where it makes sense is where the network is one signal among many. If the traces, the logs and the hosts are already here, putting the switches next to them is worth real money during an incident, because the argument about whether it is the network stops being an argument.
A network team buying this on its own is buying a corner of a large platform, and question three is not in it.
Where it wins
- A published per device price in a unit that maps to your inventory
- NetFlow retention options stated up front at 15, 30, 60 or 90 days
- A cheaper wireless access point SKU at $4 a device
- The network sits beside the traces, logs and hosts, which settles incident arguments
Where it loses
- The $7 excludes the infrastructure hosts you must buy to do the polling
- No configuration backup, so question three is not answered at any price
- On demand pricing is 45 percent above the annual rate
- A network only team is paying to stand in the corner of a much larger platform
How Datadog Network Device Monitoring scores, criterion by criterion
Datadog Network Device Monitoring pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Network devices | $7 each / mo | $16,800 at 200 devices | Billed annually. $10.20 on demand |
| Wireless access points | $4 each / mo | $9,600 at 200 APs | Billed annually. $5.76 on demand |
| Infrastructure hosts | Required | Extra | Needed to poll the devices, per the vendor FAQ |
| Cloud Network Monitoring | $5 per host / mo | Extra | A different product from device monitoring |
Domotz
Domotz publishes both of its models in full and gates nothing behind a tier. A flat $35 per collector per month covers full platform access on every device that collector monitors, or $1.50 per managed device per month in bundles of ten at $15. Unlimited users on either, no feature gating, no setup fee. That is the clearest commercial proposition on this page.
It also includes configuration backup and restore for switches, firewalls and access points, which puts a small integrator or provider within reach of question three for less than the price of a monthly lunch per site. Six of the ten tools here do not include that at any price.
The shape of the product is the limit. It is built around a collector on each network, which fits an estate of many small sites and fits one large campus much less well, and it does not go as deep on question two as the platforms above it.
If you have forty client sites of twenty devices each, this is the most efficient row in the table. If you have one site with four hundred devices, it is not.
Where it wins
- Both pricing models published in full, with no tiers and no feature gating
- Configuration backup and restore included rather than sold as a module
- Unlimited users on every plan, so adding technicians costs nothing
- A flat per site price that is easy to pass through to a client
Where it loses
- Built around a collector per network, which suits many small sites more than one big one
- Less depth on question two than the platforms above it
- A smaller supported device list than the large vendors
- Scale is the weakest score here, at 2 out of 5
How Domotz scores, criterion by criterion
Domotz pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Collector plan | $35 per collector / mo | $420 per site | Full platform access on every device it sees |
| Managed device plan | $1.50 per device / mo | $3,600 at 200 devices | Sold in bundles of 10 at $15 |
| Users | $0 | $0 | Unlimited on both plans |
| Config backup | Included | Included | Backup and restore, on switches, firewalls and APs |
Progress WhatsUp Gold
The distinguishing feature is on the pricing page rather than in the product: WhatsUp Gold still sells a perpetual license, described as a one time purchase granting the current version plus one year of updates and support.
Every SaaS product on this page has removed that option, and for an organization with capital budget and no appetite for a recurring line item, it is a real reason to shortlist.
The tiers are published and the difference between them is mostly how many installations you get and how quickly somebody answers: Business at $1,229 for 50 devices a year with 10/5 email support and one installation, Enterprise at $2,049 with two installations and 24/5 phone, Enterprise Plus at $3,469 with four.
Network Configuration Manager, which does the backups and the side by side diffs, is an add-on rather than part of any tier.
What holds the score down is the band. Every published figure is for 50 devices, and nothing above that is on the page, so a 200 device network is back to contacting sales. The product is also Windows only, which is either irrelevant or disqualifying depending on the shop.
Where it wins
- A perpetual license option, which almost nothing else in this category still offers
- Three tiers published with prices, support hours and installation counts
- Configuration management with side by side diffs and change alerts, as a module
- The tier decides how many installations you can run, which suits split sites
Where it loses
- Every published price is for 50 devices, and nothing larger is on the page
- Configuration management is an add-on rather than part of a tier
- Windows only
- The entry tier is 10/5 email support, and the lowest support score in this ranking
How Progress WhatsUp Gold scores, criterion by criterion
Progress WhatsUp Gold pricing, verified September 10, 2026
| What you buy | Price | Annual | Notes |
|---|---|---|---|
| Business | from $1,229 / yr | for 50 devices | One installation, 10/5 email support |
| Enterprise | from $2,049 / yr | for 50 devices | Up to two installations, 24/5 phone and email |
| Enterprise Plus | from $3,469 / yr | for 50 devices | Up to four installations |
| Above 50 devices | Not published | Not published | Contact sales, which is where the clarity stops |
FitWho this ranking is for, and who should skip it
This ranking is written for whoever owns the switches: an in-house network or IT lead choosing what to run, and a managed service provider choosing what to run across client sites. It assumes somewhere between one site and a few hundred devices, equipment that speaks SNMP, and at least one person who will be responsible for what the tool says.
Three shortcuts, if the whole page is more than the decision needs. If somebody on the team already knows Linux and will own the system, start with Zabbix and add Oxidized, and spend the license money on their time instead.
If you are a provider looking after networks you did not build, the configuration backup is the product and that argument ends at Auvik and Domotz, with the choice between them decided by whether you can live without a published price.
If you need an answer this week with no argument about it, PRTG is the only product here whose full price you can read and buy in the same afternoon, and the freeware tier will monitor a small office permanently for nothing.
Skip this page entirely if what you actually need is endpoint management. Servers and workstations with an agent on them are RMM, a different category with a different ranking, and the two overlap in the console rather than in what they can reach. Most providers end up running both.
The last word belongs to the failure mode, because it is not a product problem and no ranking can fix it. A monitoring system with twelve alerts people act on beats one with four hundred that they mute, and every tool on this page ships with defaults that produce the second.
Whatever you buy, the question to ask the person on call six months later is what the last alert they acted on was. If the answer takes a while to arrive, the software was never the thing to change.
ChangelogWhat changed in this update
- First published. Ten tools scored from primary sources: vendor pricing pages, product documentation and repository license files, each read on September 10, 2026 and recorded in the evidence file.
- Added what the monitoring half of this category mechanically does, the cloud based versus self-hosted split and what a collector is, and a section on why network monitoring tools and network management software return the same vendors. Capabilities read at Zabbix, LogicMonitor, Domotz, Auvik and Paessler on September 23, 2026. No price on this page changed.
FAQFrequently asked questions
What is the best network management software?
For a team with somebody to own it, Zabbix, because it costs nothing, publishes a five year support lifecycle with dates, and prices its commercial support on response time rather than on how many devices you add.
For a managed service provider, Auvik, because configuration backup is included and that is the feature that shortens outages. For a buyer who wants a published price today, PRTG.
What is the best free network management software?
Zabbix and LibreNMS, and they suit different teams. Zabbix is AGPL-3.0 and publishes an LTS every eighteen months supported for five years. LibreNMS is GPL-3.0, releases monthly, and works harder at discovering the network for you. Neither includes configuration backup, and both are commonly paired with Oxidized, which is Apache-2.0. Free of license cost is not free of labor.
How much does network monitoring software cost?
For 200 devices at published list prices, anywhere between nothing and about $19,200 a year. Domotz is roughly $3,600, PRTG about $8,904 once you apply the vendor’s own sensor conversion, Datadog $16,800, and SolarWinds $19,200 before volume discount. Zabbix and LibreNMS are free, with support for Zabbix starting at $3,900 a year regardless of size.
Why do vendors price in sensors instead of devices?
Because a sensor is one measured value and devices vary enormously in how many they have. It is defensible and it has one consequence worth knowing before you buy: the more thoroughly you monitor, the more you pay.
Paessler states the conversion itself at about 5 to 10 sensors per device, or one per switch port, so a 48 port switch monitored per port is a meaningful fraction of a small license.
Which network management tools include configuration backup?
Four of the ten here include it in what you first buy: Auvik, Domotz, LogicMonitor and, as a module, WhatsUp Gold. SolarWinds and ManageEngine sell it as a separate product called Network Configuration Manager. Zabbix and LibreNMS answer it with Oxidized alongside. PRTG and Datadog do not offer it.
Is open source network monitoring good enough for a business?
Yes, and the question is not capability. Zabbix and LibreNMS do what most commercial products do. The question is whether one named person will own the install, the upgrades and the tuning over the next three years.
Where that person exists it is the better trade. Where the plan is that somebody will find time, a commercial product is buying the availability of somebody else’s engineering.
What is the difference between network management software and RMM?
What they can reach. RMM installs an agent on servers and workstations and can therefore do almost anything to them. Network management software monitors switches, firewalls, access points and power supplies, which will never run an agent, using the protocols they already speak. Neither replaces the other and most providers run both.
Does SolarWinds NPM still exist?
Yes, as the network module of SolarWinds Observability Self-Hosted, and the product page title now says so. The practical effect is that searching for a Network Performance Monitor price returns pricing for the platform, which starts at $8 per node per month, and that NetFlow Traffic Analyzer and Network Configuration Manager remain separate purchases.
Why is Auvik ranked fourth if it wins the criterion you say matters most?
Because pricing clarity is 5 of the 40 points and Auvik scores 1 of those 5. Nothing is published: the pricing page is a form asking for a phone number and an employee count.
It scores 7 out of 7 on automation, the highest here, and those two facts together are the review. If a sales call is not an obstacle, move it up your own list.
How often should the tool poll?
Often enough to see what you need to see, and no more. Five minute polling cannot show a thirty second outage and no dashboard recovers it afterwards.
One minute polling across thousands of devices is a storage and load decision rather than a setting, which is why polling interval belongs in the sizing conversation with the vendor rather than in a feature comparison.
Do these tools need an agent on each device?
No, and that is the point of the category. Every product here monitors network equipment over SNMP and, for configuration work, SSH. Some of them also offer an agent for servers, which is optional and is a different job. If the thing you want to monitor is a switch, a firewall or a UPS, an agent was never an option.
What do network monitoring tools actually monitor?
Four things, mechanically. Interface counters, CPU, memory and temperature polled over SNMP. Reachability tested with ICMP. Traps and syslog messages the device sends unasked. And every one of those values compared against a threshold, which is what turns a number into an alert somebody sees.
Is network monitoring software the same as network management software?
Monitoring is the subset. It tells you something is down or slow. Management adds configuration backup, comparison and restore, which is the part that answers what changed. Both search terms return the same vendors here, so the distinction shows up in the quote rather than in the marketing.
Are these tools cloud based or do I host them myself?
Both models are on this list. Auvik describes itself as cloud based, with a free collector on site that proxies to the cloud. Domotz is the same shape. Zabbix, LibreNMS, OpManager and WhatsUp Gold are servers you run, and Paessler offers PRTG Hosted Monitor next to the on-premises build.
What is a collector and do I need one at every site?
A collector is a small piece of software inside the network that discovers and polls locally, then sends results to the vendor. Auvik states its collectors are free and can be deployed one per site, shared across sites, or several on one site. Domotz sells by the collector.
Do I need NetFlow to see which device is using the bandwidth?
To see who, yes. SNMP counters tell you an interface is busy; flow records exported by the device, NetFlow or sFlow, tell you which conversation filled it. That is why several vendors here sell flow analysis as a separate product rather than including it in the base license.
Where do these scores come from?
The same 40 point rubric used on every ranking here, applied to what each vendor and project publishes: pricing pages, product documentation, release histories, and the license file in each repository, all read on September 10, 2026 and dated on the page. There is no paid placement, and a vendor can dispute a fact but not a score.
Keep comparingRelated rankings
- Best Patch Management Software, and What Each One Reaches Seven tools, one 40 point rubric, every subscore published. Check the Windows Autopatch you may already own, t...
- JumpCloud Alternatives: Directory, SSO or Devices? JumpCloud is three products sold as one, identity, access and device management, and every alternative ranked...
- Ivanti Alternatives, by Which Ivanti You Are Leaving Ivanti is LANDESK, HEAT, MobileIron and Pulse Secure under one name, sold across endpoint management, IT servi...
- Atera Alternatives, and Why There Is No Half to Keep Atera is one product on one database, which is why people choose it and why leaving is all or nothing. It is a...
- ConnectWise Competitors, and the Half You Can Leave ConnectWise tops its ranking at 8.5, so nobody leaves it because it cannot do the job. They leave over price o...
- NinjaOne Competitors, and Why One of Them Is Not Leaving NinjaOne is the deepest endpoint management in its ranking, and its own site now publishes a PSA as well. So m...