MotherDuck Community Icon

Questions About Parallel Execution, Response Chunking, and Permissions in MotherduckWASM for Geospatial Data Projects

·Mar 12, 2026 05:50 PM·

Hi Guys. I have 4 questions/issues that I'd appreciate a solution or feedback if I'm doing correctly. As we move from test projects to real client projects, we noticed somethings that impact the user experience with big or geospatial datasets. Just for context, it's galileo.world, a wasm app, so there's no backend except for auth. Here they go: MotherduckWASM 1. No parallel execution - for more than one chart execution locks in - results show only when all queries are executed - It has similar behavior in the MD panel, since a query only starts after another is finished. How can I deliver query results as they return in wasm? (I'm currently using enquedQuery requests to allow cancelling dynamically) 2. Chunks in response - the chunk + wait + chunk + wait slows down bigger results such as geometries - ex: a 10mb response takes too long because the chunks are about 1mb each. Geospatial datasets when dealing with polygons, even after simplification and within restricted boundaries may have big responses. Permissions and tokens 3. Is it possible to privately share a database to a specific user/email from another organization? 4. Is it technically possible to create scaling tokens that have access only to specific databases? -> Workaround: service account + read scaling token and granting permission to it to only specific databases BUT services accounts are considered from the same org therefore have access to all org shares. Even if the share is hidden, that creates a security issue for backendless projects when I need to make publicly accessible databases to people that do not have a motherduck account. In that case our org would have to be very careful about not making org shares, only restricted permission ones. If necessary I can show it live in the prod app so you could maybe see alternatives.

Tseegi (.
·
·
Hi Patrick G.! Thanks for the great questions! I'm happy to answer them right here in the chat, but if you'd prefer to hop on a quick call to talk through everything, I can definitely connect you with a team member too. To get started, could you share your org name in MotherDuck?
Tseegi (.
·
·
1.
Queries on the same MotherDuck connection run sequentially, however, queries on different connections will run in parallel. Have you had the chance to check out connection pooling in MotherDuck? https://motherduck.com/docs/key-tasks/authenticating-and-connecting-to-motherduck/multithreading-and-parallelism/
2.
For this question, could you provide example queries where you've noticed this behavior so I can investigate it on my end?
3.
You can create hidden shares: https://motherduck.com/docs/key-tasks/sharing-data/sharing-overview/. They can be attached by anyone with the share link (in the same cloud region), including outside your org, but there's no per-email restriction available. Could you provide more detail about your use case?
4.
That's the current workaround we have, though I understand your concern. I'll ask around internally and let you know what other options we might be able to offer!
Patrick G.
·
·
Regarding 1 I mean for Motherduck Wasm - It only supports one connection and I've tried to use different instances/workers, but it seems all connected. For example, in duckdb-wasm I set as many workers as my RAM supports (usually 4 is good) and they run in parallel, creating a smoother and faster experience for the end-user. 2 is for motherduck-wasm as well. Any query that retrieves a certain amount of GEOMETRY fields becomes big in response size causing this chunk behavior to slow down. 3 if I do it creating a public hidden share, since the share link is in the project, I cannot set restriction to specific clients. Once they get the link they could do it again. I'm not sure if I was clear, but I'd appreciate a quick call with a support member, because we're using cutting edge wasm technology, so I think our use cases may help MD as well
Tseegi (.
·
·
Yeah, I think a call is helpful! Could you provide your email and org name in MotherDuck? Feel free to send it via dm!
Tseegi (.
·
·
Looping in Connor C. and Neil B. for you!
🙌1
Neil B.
·
·
Patrick G. - Great to hear from you again! Let's find some time to dig into these with appropriate resources from our side. Are you free Tuesday? I sent over an invite for 3pm ET in case that works
Patrick G.
·
·
Neil B. - works fine. I just accepted it.