Your data never leaves
It runs on your machine or your cluster, air-gapped if the site demands it. Nothing phones home, nothing is mirrored to a vendor, and the world maps keep working offline from map data you host yourself.
Why LandauOne
What makes it different
Most IoT platforms are a subscription to somebody else's cloud with a meter on every verb — ingest, store, query, view. LandauOne is software you run, and it meters nothing. Here is what that looks like in practice, and what it adds up to over ten years.
No meters, on anything
No per-sensor, per-reading, per-seat, per-camera or per-site fees. The only running cost is the hardware, and the hardware is small — see below.
Querying your own data is free
No per-query charge, no 'bytes scanned', no compute units, no retrieval fee for older months, no egress. A chart over last year costs what a chart over yesterday costs, and a colleague opening it costs nothing at all.
A decade on one drive
Ten years of ten thousand sensors at full minute resolution is about 640 GiB, summaries included — one NVMe drive, measured with the benchmark that ships in the product.
Numbers you can defend
Every reading accepted is counted, every reading discarded is counted with its reason, a redelivery is counted once, and every published benchmark figure is reproducible on your own hardware from the harness in the repository.
The questions, not just the charts
Consumption and cost under your tariff, runtime and starts, compliance evidence, drift, correlation, forecasts, sites compared, trips — answered in the console from what is already stored, with no export and no second system.
Everything in one console
Readings, bearings, positions, live cameras and logs sit side by side, replay together, and can be put next to each other at the moment that matters.
Honest when things go wrong
Two monitor modules show what arrived, what is waiting and what was lost — and why. An outage never becomes silent data loss: sources are asked to redeliver, and what genuinely could not be kept is counted, not forgotten.
Standard parts, no lock-in
History rests in a standard open-source database you can query, back up and take with you. Sign-in through your own identity provider; readings from the brokers you already run.
Ten years, ten thousand sensors
What it costs here, and what it costs metered
One fleet, one question: 10,000 sensors, one reading a minute each — 432 million readings a month, 52 billion over ten years — kept at full resolution, charted by a team that looks at it every working day. Below, the same fleet priced on four metered cloud services from their public price lists, and priced here.
List prices as read from the providers' public pricing pages on 22 and 23 August 2026, US regions, no discounts or committed-use deals. Where a provider does not publish how much disk a stored reading takes, 100 bytes per reading is assumed; where query cost depends on how often people look, two bounds are given. Your fleet will differ; what does not differ is which line items exist.
| Option | Ingest | Storage | Queries | Portal / users | Ten-year total |
|---|---|---|---|---|---|
| Amazon Timestream for LiveAnalytics | $0.50 per million writes → ≈ $26,000 | memory store $0.036/GB-h, magnetic $0.03/GB-month → ≈ $14,000 | $0.518 per compute-unit hour, 4-unit minimum: a console open eight hours a day ≈ $44,000; a wall display refreshing around the clock ≈ $181,000 | no dashboards included — bring your own | ≈ $84,000 – $221,000 |
| AWS IoT SiteWise | $1 per million messages: ≈ $5,000 batched ten readings per message, ≈ $52,000 one reading per message | hot $0.30/GB-month, warm $0.03/GB-month → ≈ $11,000; plus $0.50 per million computations for the per-interval averages ≈ $26,000 | $0.20 per GB scanned → ≈ $6,000 at a modest chart load | $10 per portal user per month: five users ≈ $6,000 | ≈ $54,000 – $101,000 |
| Azure Data Explorer, fed by IoT Hub | IoT Hub Basic B2, $50 per unit-month for 6 million messages a day: ≈ $6,000 batched ten readings per message, ≈ $18,000 one reading per message | blob storage $0.0208/GB-month → ≈ $6,500; the SSD hot cache is the cluster’s own disk | no per-query meter — the cluster is the meter, and it runs whether anyone is looking or not: the smallest production cluster (two E2ads v5 engine nodes at $0.131/h plus the $0.11 per vCore-hour Data Explorer markup, and two D3 v2 data-management nodes at $0.293/h) ≈ $1.29 an hour → ≈ $113,000 | Data Explorer dashboards included; Azure Managed Grafana ≈ $25 a month plus $6 per user: five users ≈ $6,600 | ≈ $126,000 – $144,000 — or ≈ $29,000 on the single dev/test node, which has no SLA |
| InfluxDB Cloud Serverless | $0.0025 per MB written → ≈ $10,000 | $0.002 per GB-hour → ≈ $46,000 — and the plan caps retention at one year, so ten years is not available on it at all | $0.012 per hundred queries → under $1,000 | no dashboards included | ≈ $57,000, for at most one year kept |
| LandauOne on a rented dedicated server | included | included — 1 TB NVMe is standard in this class | included, unlimited | included, unlimited users | ≈ €7,000 – €14,000 (a 64 GB, dual-NVMe dedicated server lists at €60–120 a month) |
| LandauOne on your own server | included | included | included, unlimited | included, unlimited users | ≈ €13,000 (a €3,000 machine, replaced once, plus ≈ €650 a year of electricity) |
| LandauOne on a hyperscaler VM | included | 1 TB block storage ≈ $90 a month | included, unlimited | included, unlimited users | ≈ $44,000 (an 8-vCPU, 32 GB instance ≈ $280 a month plus the disk) — the dearest way to run it, and still no meter |
Storage is never what decides a metered bill — it is 5–20 % of every line above. What decides it is usage, in one of three shapes: paying to look (Timestream bills compute-unit hours while a dashboard refreshes), paying for the maths (SiteWise bills every average it computes), or paying for the capacity to look (Data Explorer rents a cluster by the hour whether anyone is watching or not). All three grow with how useful the system becomes, not with how many sensors you have. LandauOne precomputes the maths once, and looking is included — the self-hosted totals are the whole bill.
Reading your own data
No charge for querying what is yours — as it should be
A metered platform prices every look: bytes scanned, compute-unit hours, retrieval from a colder tier, rows returned, seats at the screen. LandauOne treats the opposite as the baseline.
Old months cost what new ones cost
There is no cold tier you pay to rehydrate and wait for. A month from three years ago is as queryable, and as fast to summarise, as last month — the history is one store, not a cheap archive behind a retrieval fee.
The maths is already done
Averages, medians, percentiles, spike envelopes, consumption and coverage are computed as the data settles, once, for everyone. Asking for a year's P95 is a read, not a scan you are billed for by the gigabyte.
As many eyes as you like
No per-seat or per-viewer pricing, no egress. A dashboard shared with the whole shift costs exactly what it cost with one person watching.
The numbers behind the storage and throughput claims are on the benchmark page, with the harness that produces them. See the benchmark