2 ms·
I'm legitimately curious if anyone has any clues as to how they handled this with cloud providers. Technically it doesn't seem that hard -- just spin up a ton
by crazygringo 7y ago
I'm legitimately curious if anyone has any clues as to how they handled this with cloud providers.
Technically it doesn't seem that hard -- just spin up a ton more servers. Unlike Facebook or Twitter or Reddit there isn't massive communication between all users at all times -- just one-off video calls that are "relatively" trivial to distribute among servers. (Sure there's cloud recording and stuff too, but the point still stands that this is certainly among the way easier products to scale.)
But assuming they're using cloud providers, I'm curious what percentage of spare cloud capacity they've wound up taking? If they've had to split up traffic between multiple major providers just to handle it?
Or if the increase is well within the capacity of any single cloud provider to handle easily, e.g. massive daytime spikes are essentially just using the same servers Netflix uses at night?
- StandardFuture 7y agoMaybe we can assume if people are on Zoom then they are also not on Netflix? But yes, would be very interesting to get a post-survival writeup.
- jabart 7y agoThe public zoom filling showed Zoom had a lot of co-located servers. From their SEC filing. "We currently serve our users from 13 co-located data centers in Australia, Brazil, Canada, China, Germany, India, Japan, the Netherlands and the United States. We also utilize Amazon Web Services and Microsoft Azure for the hosting of certain critical aspects of our business. "