The watchman's job.
Argusnodes is named after a watchman who could not be put to sleep, and it started in a one-bedroom flat a long way from anywhere. This page is the name, the mark, the method, and the things we refuse to do.
A hundred eyes, and never all of them shut.
In the Greek myth, Argus Panoptes ("all-seeing") had a hundred eyes across his body. When some slept, the rest stayed open, which made him the one guard in the world who could not be snuck past. Hera gave him her hardest job: watch Io, day and night, and let nobody near.
He was finally beaten by a story. Hermes sat down beside him, played the pipes, and talked until the last eye closed. Hera, in mourning, set his eyes into the tail of the peacock, where they still are.
We took the name for the first half and the warning from the second. A node is something that has to be watched all the time by someone who cannot be lulled. Dashboards that say green while p99 climbs, status pages that stay up while the node falls over, a "dedicated" plan that turns out to share a machine: those are the pipes. The job is to keep some eyes open regardless, and to report what they see, not what anyone would like to hear.
A one-bedroom flat, a long way from a hub.
Argusnodes did not start in a datacentre, an accelerator, or a city you would guess. It started in a one-bedroom flat (a 1BHK, if you know the term) in a remote corner of Asia, far from the places where infrastructure companies usually begin.
That matters more than it sounds. If you have ever tried to run serious infrastructure from far away, you already know what it is to be rate-limited, metered, and told that the thing you need is "enterprise only". We built the option we wanted to buy: the whole machine, every capability on, one flat price, and proof instead of promises. Nothing about that needs a head office. It needs a decision, made in advance, never to charge for things a machine already does.
How a quote happens in an hour.
The hour is not a slogan. It is possible because most of the work is done before you ask.
You send the workload
Chain, node type, and what you will throw at it: bursts, backfills, subscriptions. Three fields on the contact page or any chain page.
We pull the runbook
Every researched chain already has one: clients, interfaces, sync method, storage, all taken from the chain's official documentation. New chain? We research it first, then quote.
Spec and flat price back
One monthly figure from five public bands, hardware named, city chosen, nothing metered. If the workload needs a bigger machine, the quote says so and why.
Build, then watch
The node is provisioned, synced and tuned for the workload, then watched: live p50 and p99 in your dashboard, humans on call 24/7.
Unknown stays blank.
Every chain page on this site is built from two kinds of data, and we label which is which. Where the official documentation does not say, the field is left empty. We would rather show you a dash than a guess.
Ten things we will not do.
- 01Rate-limit youIf the machine can serve it, you can send it.
- 02Call a shared cluster "dedicated"One tenant per machine. Every time.
- 03Charge for trace, debug or WebSocketsThey are things the software already does.
- 04Put support behind a paywallPriority support is the only kind we have.
- 05Inflate chain countsWe say how many are researched and how many come from the registry. Separately.
- 06Invent a hardware specEvery figure on a chain page has a source link. Unknown stays blank.
- 07Print customer logos we have not earnedOr team photos we have not taken. Placeholders are lies with nicer fonts.
- 08Quote you next weekA firm flat price inside the hour, around the clock.
- 09Pretend the price is the whole storyLive p50 and p99 for your node are in your dashboard from day one.
- 10Fall asleep to a good storySee the myth. That was the watchman's one mistake.
Why the site looks like an engineering drawing.
Because the product is one: verifiable hardware, measured latency, a spec you can inspect. Every device on these pages is borrowed from drafting, and each one encodes something true about the content.
| Convention | On a drawing | On this site |
|---|---|---|
| Title block | The boxed strip under a drawing that says who drew it, when, and to what revision. | Under every figure and table: where the facts came from and when they were checked. |
| FIG. 01, 02, 03 | Numbered figures, so a drawing can be cited precisely. | Only real figures get a number: the latency chart, the configurator, the red-line ledger, this logo. |
| Dimension line | A thin line with arrowheads giving a measured size. | The stat strips: SLA, cities, uplink, rate limit. Measured things only. |
| Red-line | A correction marked in red ink on a drawing. | The struck-through add-on fees on the pricing page, corrected to zero. |
| REV B | The revision letter. Every change gets a new one. | The design is on its second revision. Rev A was rejected for looking like everyone else. |
| Sheet | One drawing in a set. | Every page is a sheet on the same grid paper. The 404 is "sheet not found". |
Built on order, not on inventory.
Shared providers must pre-run every chain they list, so their coverage stops where their capital stops, and their meters exist to protect overloaded clusters. We provision each node when it is ordered. No idle fleet to subsidise means no meter needed to protect it, and a chain nobody else will run costs us a runbook, not a datacentre. The full argument, line by line, is on the Why Argusnodes page.
Judge us on the numbers.
Your node's live p50 and p99 are in the dashboard from the first hour. That is the point of it.