IoT Platform vs LoRaWAN Network Server
IoT Platform vs LoRaWAN Network Server
A LoRaWAN network server handles the radio side: it authenticates devices, deduplicates uplinks arriving through several gateways, manages data rates, and hands you a JSON payload. An IoT platform starts where that stops. It decodes the payload into named readings, stores the history, draws the dashboards, evaluates the alert rules, and controls who sees what. Most deployments need both. Many organizations buy only the first and discover the gap later.
The confusion is understandable, because the two are often sold together and the boundary between them is invisible from the outside. It becomes visible on the day a working deployment delivers raw bytes to somebody who expected a dashboard.
Neither piece is optional in a working deployment. They simply do different jobs, and knowing which does what is the difference between a sensible architecture and an expensive surprise.
|
Job |
LoRaWAN network server |
IoT platform |
|
Authenticating a device onto the network |
Yes |
No |
|
Removing duplicates when several gateways hear one message |
Yes |
No |
|
Managing data rates and radio parameters |
Yes |
No |
|
Queuing a downlink for delivery to a device |
Yes |
Decides when and why to send one |
|
Turning a payload into named readings |
No |
Yes |
|
Storing history for months |
Rarely, and briefly |
Yes |
|
Dashboards |
No |
Yes |
|
Alert rules and escalation |
No |
Yes |
|
Users, roles and who can see what |
Basic, operator focused |
Yes, including customer-facing access |
|
Separating one customer from another |
No |
Yes |
|
Billing, plans and commercial packaging |
No |
Yes |
Some vendors call the second column an application server rather than a platform. The words vary; the split does not.
Plenty of deployments run on a network server alone, and for some of them that is exactly right. ChirpStack is open source and widely deployed. The Things Stack is available both as a hosted service and to run yourself. Both do the radio job properly.
What you get is devices joining, uplinks arriving deduplicated, downlinks queued, and a payload delivered to wherever you point it. What you do not get is meaning. The payload is bytes, and turning those bytes into a temperature that somebody can chart is your problem. So is storing them, drawing anything, deciding when a reading is worth waking somebody up about, and controlling which of your customers can see which devices.
That is a reasonable place to stop if you have engineers, a small number of device types, and somewhere else for the data to go. It stops being reasonable at about the point where somebody who is not an engineer needs to look at the data, or where a second customer appears.
The reverse case is less common and it catches people out just as often. Platforms such as ThingsBoard handle devices, dashboards and rules, and they are not LoRaWAN network servers. They expect something else to have already done the radio work.
If your devices are cellular or wired, that is fine, because there is no radio layer to run. If they are LoRaWAN, you still need a network server underneath, whether you run it, rent it, or receive it as part of something else. Buying a platform does not remove that requirement; it just makes it somebody's job to notice.
In practice the join between them is smaller than people expect, and it happens on the gateway rather than in software.
|
Step |
What happens |
|
The gateway |
Runs a packet forwarder, most commonly the Semtech UDP forwarder, and is configured with the address and port of a network server. That single setting is the boundary. |
|
The network server |
Receives the traffic, authenticates the devices, discards duplicates, manages the radio parameters, and passes the payload on. |
|
The platform |
Receives the payload, decodes it into named fields, and everything else follows from there. |
That is why the two are so often bundled: once a gateway is pointed at a network server, whoever runs that server is well placed to hand the payload straight to a platform. It is also why the split is invisible until something goes wrong, at which point knowing which side of the line a problem sits on saves a great deal of time.
Three questions settle it for most people.
|
Question |
What the answer tells you |
|
Does anybody who is not an engineer need to look at this data? |
If yes, you need a platform. Raw payloads and a network server console are not something you can hand to an operations team, and building a viewer for them costs more than it looks. |
|
Will more than one organization use it? |
If yes, you need a platform. Separating customers, giving each their own view, and controlling who sees what a network server’s job is not and never has been. |
|
Are your devices LoRaWAN? |
If yes, you need a network server, whether you run it, rent it, or get it as part of a platform. If your devices are cellular or wired, you do not. |
Most people asking this question answer yes to at least two, which is why most real deployments end up with both.
For completeness
Where ioX-Pulse and ioX-Bridge Sit
For completeness, since this is our site: ioX-Pulse is the platform, and ioX-Bridge is the LoRaWAN network server underneath it. They are separate products in the ioX-Connect suite, and either can be used with something else in place of the other. Most people running ioX-Pulse never need to touch the network server directly, which is the arrangement this page has been describing throughout.
ioX-Pulse
ioX-Bridge
Answers to the questions we get most
Please reach out to us at: sales@iox-connect.com if you have any additional questions that are not addressed below.
Latest blog posts
Check out our recent blog posts for everything IoT related
- ioX-Connect
- April 18, 2026
- Ockert Fourie
- March 26, 2026
- Ockert Fourie
- October 10, 2025
