7 ms of latency is nothing but a number because I can give you a thousand examples of hitting various services way closer than that and having 30+ ms of latency. Real world experience matters and you basically calling me a liar for bringing up actual issues with streaming is insulting. Overwatch is an online game and so are the fighting games you're talking about. It's clear that while you have no issue insulting others, you don't feel bad about being disingenuous yourself.
I have been clear that I don't want games built solely around streaming that take away from the games we're getting today. If you have a different opinion than that fine, but that opinion on its own is not based on pseudo science nor does it lack merit.
Obviously my data center example is pure hyperbole, you have still not offered a solution to the problem however. Concentrating on routing is all well and good, but that problem is incredibly immense and involves negotiating with various ISPs, government entities, etc.
the number is the speed of light every 1000 miles in latency. Which you should know if you're gonna bring up a point about getting around the light barrier in networking.
I didn't call you a liar you can be wrong and not lying.
I am relating real world experience I'm just saying why people have crap latency that could be fixed and why more often than not it's not getting fixed or addressed properly.Whether it's a game or streaming service the packet ends up udp or tcp and how it get managed before it hits your router, how it's managed in your router and how your gateway to your isp manage it is just as vital and more so then a bad hop you may see at a data center or beyond.
You can be clear all you like doesn't remove you from having an opinion that is off or misguided.
I offered various solutions in cloud streaming topics already but here are some basics.
1. Considering it's clould I'd want the linux kernel to be up to date or at least 4.1 or higher.
2. All queueing mechanism use cake with diffserv
3. Traffic prioritity is used and built for diffserv4 or higher markings
4. Private peering for routing. Private peering literally skips most routing problems you're bringing up you'd know if you studied what vpn brings to the table.
5. Offer devices that do low latency well and aren't lip service. Most gamers will need proper aqm be it stock or custom firmware to deal with latency eventually.
6. Interrupt Coaslecing on all hosting devices would be minimum needed to keep latency in check
7. TXQueuelen on certain interfaces would need to be more than 1000 same for other settings related to how many packets are handled per interface or cycle.
8. Napi Polling would be used.
You're feelings are irrelevants to certain facts being laid out. Current infrastructure is more of joke due to software and tuning than hardware.
