Kumulus documentation
A distributed cloud built from machines other people own — and the three ideas that make that work.
Kumulus is a cloud whose compute comes from machines contributed by independent providers rather than from datacentres we own. That one difference is behind almost everything else on this site: providers are not interchangeable, they are not always available, and the platform has to be honest about both.
Three ideas carry most of the weight.
Start here
Provider tiers — Flare, Stars and Tutors. What kind of machine sits in each, and why a tier is a badge a machine earns rather than a separate product.
MOSAIK — the vocabulary for what a given node can actually do, so work is matched to machines that can run it instead of failing halfway.
BIDEEW — how the platform chooses which provider runs a given workload, and why it would rather refuse than guess.
Read in that order if you are new: tiers describe the machines, MOSAIK describes what they can do, BIDEEW decides where work goes.
Contributing a machine
Joining as a provider walks through the shape of onboarding a node — four steps, and what each is for.
What lives where
This site covers Kumulus: how the network is organised and how to take part in it.
The wider Kollectyve Network — its governance, the consortium model and the litepaper — is documented at docs.kollectyve.com. Concepts are explained there once rather than restated here.
Getting help
Questions and problems go to Discord.