I find this surprising:
> As a small nonprofit organization, we cannot afford to purchase all of the physical computers that are necessary to support everyone who relies on Signal while also placing them in independent data centers around the world. Only a select few of the very largest companies globally are still capable of doing this.
Signal may be “small,” but they’re spending plenty on this. Registration is expensive and hard to do without using one of the large expensive providers. But there’s $7M for servers, storage and bandwidth. These are comparatively easy: servers and storage (especially for a service like this where availability for the substantial majority of the data is not terribly important) come in nice pre-manufactured boxes that can easily saturate 10Gbps and can store quite a few TB at very very high IOPS. [0]. And the forwarding model isn’t very latency sensitive - several hundred ms for most users is fine, and sending media via Signal is quite slow regardless. So having many points of presence doesn’t seem terribly important. I bet that two small colocated facilities could cover all of North America quite nicely.
Bandwidth costs outside the cloud world, at least in North America, are comically cheap compared to the major clouds.
[0] A service like Signal ought to need relatively little processing compared to bandwidth and storage for the data plane. AWS and the like may not have a particular good match in their catalog for this use case.