> (A benefit that's not talked about, that probably doesn't matter for Uber, but may matter for smaller players, is that the driver phone plan is subsidizing bandwidth costs by not having to pay as much for inter-datacenter bandwidth.)
Is the amount of data being transferred per trip significant? I mean, how much of a difference would it make for "smaller players" (Lyft?) to go this route? And if the savings are significant, does that mean that the cost to drivers is non-significant? If Uber has enough money that the savings on datacenters to them isn't huge, I'd hope they would eat the cost if the phone-as-a-datacenter route ends up being a burden to their individual drivers.
Are they doing dual writes from the phone? That is, updating datacenter A and B (and possibly C) at the same time? If so, then yeah this will cut down on their need for backhauling between datacenters. If they are only using the phone as the arbiter of truth during a datacenter failover, then no they're not saving any money. You're going to have a circuit or transit in place and that has a monthly fee at 95/5 or flat rate, not per byte transferred.
Is the amount of data being transferred per trip significant? I mean, how much of a difference would it make for "smaller players" (Lyft?) to go this route? And if the savings are significant, does that mean that the cost to drivers is non-significant? If Uber has enough money that the savings on datacenters to them isn't huge, I'd hope they would eat the cost if the phone-as-a-datacenter route ends up being a burden to their individual drivers.