Straight answers.
What you pay for, and what you never do.
What exactly is a "server"?
One machine running Trinium. It runs one pipeline at a time for unlimited people, pipelines, runs, rows and cores, so a bigger machine never costs more. Each server you add lets one more pipeline run at the same time.
A deployment is a separate installation with its own Postgres database, such as production plus disaster recovery, or a second business unit. Staging and production both live inside one deployment. Pro covers one deployment with as many servers as you buy; Enterprise covers unlimited servers and deployments.
Why per server, and not per user?
Because a server is a machine and a user is a person, and only one of those should decide who in your company is allowed to build a pipeline. Pricing per person rations the work; pricing per machine does not.
It also keeps the bill predictable: it changes when you add a server, and never when you hire.
How many people can use it?
As many as you like, on either edition. Every account is free, whether that person builds pipelines, runs them or only reads the results, and no edition counts pipelines, runs or run history.
How is it billed?
Prices are quoted per month and invoiced once a year, in advance. You pay by bank transfer or against a purchase order, and your price is fixed for the length of your contract; two- and three-year terms keep it fixed that long.
Adding a Pro server or moving to Enterprise partway through a term is charged pro rata for the rest of it, so everything renews on one date.
What support is included?
Every licence includes support from the engineers who build Trinium. Pro is answered in business hours. Enterprise gets a same-day response and a named contact, and an Enterprise agreement can add a response-time SLA.
How do I move from Pro to Enterprise?
Ask us for an Enterprise key, paste it into your environment file in place of the old one, and restart. It is the same image and the same database, so nothing is migrated and the only downtime is the restart. Partway through a contract, you pay only the difference for the rest of the term.
Renewals, and moving back to Pro, are the same key swap.
What happens when a licence lapses?
New runs are refused. That is all. Nothing is deleted and nobody is locked out: people still sign in, open and edit pipelines, and browse run history. A run already in progress finishes, the refusal names the reason, and schedules pause, then pick up again as soon as the licence is renewed.
You get notice first: a banner on every page for the last seven days before a licence expires, and Settings always shows the exact date. A lapse never switches off anything already configured, so single sign-on keeps working. Renewing is a new key and a restart.
What it takes to run, and how far it goes.
What do I run it on?
A Linux server with Docker Engine 25 or newer and Compose v2.20 or newer, on x86-64, with at least 32 GB of memory and 100 GB of disk, which is the size the engine is certified at. A virtual machine in any cloud or in your own data centre is fine. The default install brings its own Postgres; if you would rather use yours, it needs PostgreSQL 14 or newer.
Production runs on Linux; macOS and Windows are for evaluation. The full install guide arrives with your trial key.
Do I need a DBA, a Kubernetes cluster or a platform team?
No. The default install is three commands on one Linux server, and it brings its own Postgres. There is one value to fill in: your licence key. Trinium generates its own encryption key on first start and sizes its memory from the machine it runs on.
It also looks after itself. It backs itself up on the schedule you set, to local disk or, on Enterprise, to cloud storage, so there are no database scripts to maintain. HTTPS needs no reverse proxy: put a certificate and key in its folder and restart, and renewed certificates are picked up without a restart.
Will it handle our volume, and how do I size it?
It is certified at ten million rows: a 24-node pipeline ran ten times in a row on one 8-vCPU, 32 GB server, with every row accounted for on every run. Larger runs have also completed on that machine, up to 500 million rows through an 80-node pipeline. Rows stream through in batches, and sorts and joins spill to disk when memory runs short.
Execution is single-server and one pipeline at a time. Every node in a pipeline runs at once, so one pipeline uses the whole machine, and more servers pointed at the same Postgres and shared object storage run more pipelines side by side. Every run records its peak memory, which tells you when you need a bigger server, and its queue wait, which tells you when you need another one. A bigger server never changes the price.
Plan one more server if analysts will test pipelines while long overnight jobs are still running, because a test run waits for the server like any other run.
Where your data goes, and what happens to it.
Is my data ever sent to you?
No. Everything runs on your own servers, and your data stays in your own databases and storage. We run no service on your behalf, collect no telemetry and have no way to see what you run. The install needs the internet only to pull the Docker images.
Does Trinium phone home?
No. Your licence is a signed key that Trinium checks offline, against a public key built into the product, so there is no licence server and no usage reporting.
Trinium connects only to what you configure: the sources and destinations in your pipelines, your mail server and alert webhooks, your identity provider, and your backup storage.
Does Trinium use AI?
No, and nothing in it sends your data to an AI service. Every step in a pipeline is one somebody configured and anybody can read, and nothing decides for itself what to do with your data.
If you want a model in a pipeline, the Python Script node on Enterprise runs your own code, and libraries such as scikit-learn can be added to its image.
Are we locked in?
Your data is not. It never leaves your own databases and storage, and everything Trinium writes lands in ordinary formats: tables in your own database or warehouse, files such as CSV, Parquet and Excel, and Iceberg tables. Your pipelines are stored in your own Postgres database too.
The licence is checked offline, so an instance keeps running for its whole term without ever contacting us, and source-code escrow can be written into an Enterprise agreement.
Thirty days of the whole thing.
Do I need a credit card?
No. There is no card on file and nothing converts to a paid plan when the trial ends. New runs stop until you buy, and everything you built stays where it is.
Can I trial it on my laptop?
Yes, with Docker Desktop on macOS or Windows. Trinium sizes itself from the memory Docker Desktop is allowed, so give it as much as you can spare. For anything you want to show a colleague, use a Linux server, so the address works for both of you.
Will my work survive if I buy Pro rather than Enterprise?
Yes. Your paid key replaces the trial key on the same installation, and everything you built stays. A pipeline that uses an Enterprise-only node, such as Python Script or Apache Iceberg, stays saved and editable; only starting a new run of it is refused.
Can we extend the trial?
Usually, yes. Tell us what is left to evaluate and we will send a new key.
Coming from another tool.
Can we bring our workflows from another tool?
Not automatically. Workflows built in other tools lean on steps and macros that do not translate one to one, so moving means rebuilding them on the canvas, and we help you rebuild your first pipelines during the trial.
Start with the pipeline that matters most, and see how long it takes.
Still have a question?
Ask it, and the people who built Trinium will answer.