Engineering guides / Realtime communication and P2P engineering
By Arte Soft engineering · 24 October 2026
Group calls: compare SFU and peer-to-peer topology before scaling
The topology that works for two participants may be unsuitable for a larger room. Compare peer-to-peer mesh and selective forwarding around the actual meeting behavior, rather than treating one as universally cheaper or more private.
Describe the room, not only the headcount
A class with one presenter and many listeners differs from a small meeting where everyone speaks and shares video. List active cameras, screen sharing, expected device types and recording needs. Include the weakest network connection you intend to support. Participant count is important, but the number and quality of simultaneous streams can be more informative when choosing a topology.
Understand the mesh trade-off
In a peer-to-peer mesh, each participant can need connections and outgoing media for multiple others. That can increase upload and device demands as the room grows. Relay may still be required for some paths. A direct-path demonstration with a few well-connected laptops does not establish that a similar design will work on lower-powered phones or constrained mobile uploads in the intended audience.
Evaluate an SFU as infrastructure
A selective forwarding unit changes how streams travel and introduces a media-service operating requirement. Assess geography, routing, availability, capacity and monitoring. Determine whether clients can subscribe to different qualities and how active speakers are selected. Do not reduce the comparison to a server bill alone; client battery, support complexity and the cost of failed meetings also affect the product’s business case.
Make privacy requirements explicit
Transport encryption and content confidentiality across the chosen infrastructure need separate analysis. If end-to-end media protection is required, verify the supported mechanisms and the effects on recording or server-side processing. An SFU does not automatically imply that every proposed confidentiality property is impossible, and P2P does not automatically solve identity, recording or endpoint compromise. Document the boundary rather than relying on architecture labels.
Run a comparable room experiment
Use the same devices, network limits, camera activity and meeting script for each option. Measure upload, CPU, thermal behavior, quality changes and recovery when a participant’s network worsens. Include the moment a screen is shared and several people begin talking. Report the tested range and failure conditions, not a maximum room size extrapolated from an idealized local benchmark.
Topology experiment brief
- Specify active cameras and speakers, screen-sharing behavior and the weakest supported device before comparing room architectures.
- Measure client upload and resource use alongside server traffic, since moving cost between components does not necessarily remove it.
- Agree the point at which observed room behavior would justify changing topology, rather than extrapolating an unlimited growth promise from a small demonstration.
Buy a decision before a platform migration
A short assessment can produce a topology recommendation, deployment outline and evidence-based migration boundary. Arte Soft’s Reel Screen Recorder is a published media product reference, while Wonek chat and call engineering should be identified as R&D. Neither substitutes for a client-specific room test. The deliverable should explain why the selected approach fits this audience, which assumptions remain uncertain and which growth signal would trigger another architecture review.
Relevant Arte Soft work
Wonek — communication engineering R&D, not a shipped or audited messaging product is our own product or engineering project. It provides context for the work we discuss here; it is not a claim of identical results for every client deployment.
Sources and further reading
Need help with Group media topology assessment?
Send the current app, system or brief, the behaviour you need and your constraints. We can agree a discovery, implementation or repair scope with concrete acceptance criteria.
Discuss this with Arte Soft