Why we built a modelling GIS that has no server
· geonodal·map
Most modelling tools that moved to the web did it by moving the computation to someone else’s computer. You upload a model, a server runs it, and results come back. That works, and it also creates a question every utility has to answer: where did the network model go, and who can read it?
We took the other route. The solvers, the database and the Python runtime are all compiled to WebAssembly and run inside the browser tab. There is no upload step because there is no server to upload to.
What that buys
No queue. A run starts when you click it, not when a worker frees up.
No trust problem. The honest answer to “where does this data go?” is “nowhere”.
No install. Which, for anyone who has fought a corporate SOE to get a modelling package approved, is not a small thing.
What it costs
Everything is bounded by the machine in front of you. A very large model is limited by your RAM instead of by a server tier. We think that is the right trade for most day-to-day work, and we are explicit about it rather than hiding it.
If you want to follow along, the forum is public and you can vote on what lands next.