For real time communication (like calls), you need global data centers. Hyperscalers aren’t required but make it easier to deploy globally since you rely only on a single provider.
Unless things have changed, calls on signal between you and people on your contact list are p2p. This is necessary to have a call experience like what people are used to. The latency of going through a relay makes conversation more difficult. Yes that does mean that someone on the network can see that there is a signal call between the 2 ip addresses (but they wouldn’t know the identities of the users) https://signal.org/blog/signal-video-calls/
For chat alone they are right.
For real time communication (like calls), you need global data centers. Hyperscalers aren’t required but make it easier to deploy globally since you rely only on a single provider.
Why wouldn’t P2P work for real time comms?
Could this violate Signal’s security model?
Your country would be able to determine exactly with whom you’re speaking by nature of IP’s being public.
It’s arguably more difficult if there’s a server intermediary, especially when speaking to someone in a different country.
Unless things have changed, calls on signal between you and people on your contact list are p2p. This is necessary to have a call experience like what people are used to. The latency of going through a relay makes conversation more difficult. Yes that does mean that someone on the network can see that there is a signal call between the 2 ip addresses (but they wouldn’t know the identities of the users) https://signal.org/blog/signal-video-calls/
Ah, alright. Fair enough.
Though in that case my earlier assumption about AWS doesn’t hold true either.
Signal has the option to relay calls, but it is off by default.