3 ms·
I'm curious how much traffic he's sending (it doesn't sound bandwidth intensive): > Mr. Howse spent weeks finding server space in regions where demand bubbled
by boulos 10y ago
I'm curious how much traffic he's sending (it doesn't sound bandwidth intensive):
> Mr. Howse spent weeks finding server space in regions where demand bubbled up. He is trying to save money by avoiding cloud services from the likes of Amazon Inc. or Alphabet Inc. “It’s incredibly expensive because of the amount of bandwidth this game uses,” he said.
[Edit with more quotes]
> Three months ago, Steven Howse struggled to pay rent. Now, the 32-year-old developer is trying to keep his hit videogame running smoothly as it pulls in more than $100,000 in revenue daily.
> “Slither.io” is profitable, Mr. Howse said. He pays about $15,000 monthly for online-hosting services, and shares revenue with Apple Inc. and Google.
So $3M/month in revenue on $15k/month in hosting ;). I'm sure much of the $100k/day is recent and likely a fad, but I guess I'm surprised that he's apparently spent weeks trying to get capacity while his revenue is skyrocketing "just" to keep his bandwidth bill down. (It seems like all the networking would be effectively player state back and forth, not assets).
Disclosure: I work at Google on Cloud.
- vthallam 10y ago>I'm sure much of the $100k/day is recent This is probably the reason and the unexpected scale. Initially as a broke(referring to part where struggled to pay rent) 32 year old developer, you don't want to share spend a lot on AWS/GCP. If the revenue stream is consistent, I'm sure, he is going to move to either of the platforms. So much less headache that way.
- ebsalberto 10y agoWrong. He isn't even running on cloud anyway. For this type of workload, bare-metal is 10 times a better option, both in terms of performance and costs.
- mullen 10y agoThe "problem" this dev is having is what AWS was built for. He could get his OS of choice, customize it with his server software, then copy the AMI to various regions, spin up dedicated instances as he needed them, use what he needed in various regions and direct players to their closest server. Turn on and turn off servers as need arises and they are close to players. After his game settles down to predictable traffic, he can move to some bare cost ISP. He would not have to call anyone and within a hour or so he could have a game that gave a much better experience. What kills a flavor of the month game like this is being laggy and having an awful experience. The only thing I think would be an issue is the bandwidth charges. I wonder why his game creates so much traffic?
- vthallam 10y agoThe game is very latency sensitive and it makes complete sense to use a cloud and distribute the servers based on the user load : locations and that's what he plans to do as mentioned in the article.
- ben_jones 10y agoIt seems popular for multiplayer games to manage state by taking all user input and processing it on the game server. With the prevalence of cheat clients this makes the most sense, and I've seen some games do it to limit state information easily accessible on the client (to combat botting tactics). Could contribute to large bandwidth usage if allowed to run rampant (which is probably what happens in blitz-scaling situations). EDIT: After playing the game it seems plausible that each and every mouse event is sent to the server. That's a low of mouse events.
- boulos 10y agoBut you only pay for responses (ingress is free, you pay for egress). Can someone playing the game see how much traffic they're receiving to make this discussion more informed?
- x0x0 10y agoslither is similar to a fps, so it's very very latency sensitive. It's frankly often unplayable at peak times on the W coast on xfinity.