[Data Center USA] Ping increased. Report your changes here.

[Namingaway]

routing tables are still being rebuilt ripple by ripple. It will likely take weeks for everything to settle down from the changes. I bet that the DNS changes haven't fully propagated yet either, Though I'm sure the main root DNS servers are updated otherwise it wouldn't work at all.

Regarding the data center location, unless SE has confirmed a state for it, I wouldn't make any assumptions until things settle in terms of routing.

Ahhh ok thanks!

[Raikki]

Ahhh ok thanks!

Unfortunately that person doesn't quite know what they're talking about. It's possible that NTT could improve their internal routing in the future, but the location of the datacenters is set in stone now. The Colorado address you're seeing is a business address that NTT America's IP's are registered to, but is not where they're being used. This is abundantly clear if you simply look at their website:

http://www.us.ntt.com/en/services/da...center/us.html

They are not running a Colorado datacenter. They have facilities in California, Virginia, and Texas. No matter what conspiracies you want to entertain, even if you want to ignore the fact that the route terminates right after a clearly-labeled "sacramento, ca" hop, the fact is that the low pings northwest players are getting now would not even be possible if it were in Virginia or Texas no matter how good the routing was. You're being asked to believe the Earth might be flat.

[Kosmos992k]

Unfortunately that person doesn't quite know what they're talking about. It's possible that NTT could improve their internal routing in the future, but the location of the datacenters is set in stone now. The Colorado address you're seeing is a business address that NTT America's IP's are registered to, but is not where they're being used. This is abundantly clear if you simply look at their website:

http://www.us.ntt.com/en/services/da...center/us.html

They are not running a Colorado datacenter. They have facilities in California, Virginia, and Texas. No matter what conspiracies you want to entertain, even if you want to ignore the fact that the route terminates right after a clearly-labeled "sacramento, ca" hop, the fact is that the low pings northwest players are getting now would not even be possible if it were in Virginia or Texas no matter how good the routing was. You're being asked to believe the Earth might be flat.

Insulting people doesn't help your accuracy.

What I stated about routing tables is correct, it's a matter of fact that routing tables are dynamic by design and continue to optimize where possible. I made no statement about any specific data center location, deferring to better information should SE release it. NTT improving their internal routing is one thing, but routing into NTT can optimize regardless of what NTT do internally.

[Raikki]

Insulting people doesn't help your accuracy.

It's not an insult to warn people away from misinformation about "maybe they aren't really in California." As for being able to fix routing issues regardless of NTT, that's not entirely true either because of the following: NTT is not advertising that prefix externally. That means only NTT's internal routing knows how to reach it. Instead, NTT only advertises the much broader address space of 204.0.0.0/14. Assignments within that space are being used in geographically disparate locations. In order for this to work, routing to an address in that space simply tries to find the best way to get you onto NTT's network as close to you as possible. For me, this is in NYC. From there, NTT's network is responsible for getting me all the way to California.

The only thing outside of NTT's power is improving routing up to the point where you get on NTT's network. If this is where someone's issue is, then sure that can be fixed without NTT. But if your issue is in how NTT is routing you internally, then no that won't get fixed without them, and the only other recourse is to use VPN services to forcibly change where you get on their network.

None of that is to suggest a doom and gloom scenario, though. It's quite possible that things will improve for some people. There isn't really much that can be done for Europeans or South Americans, though, considering even ideal routing is going to get them fairly high latency to this location.

[Kosmos992k]

It's not an insult to warn people away from misinformation about**snip**

Good, then perhaps you should be warning them away from yourself.

I just tracrt'd Behemoth from my corporate network, I get a RTT averaging 67ms, which includes all the hops through NTT. From my residential connection, my RTT is more than twice that. That's not NTT's fault it's my ISP and their crap routing. The route from my corporate network hits an external ISP after 5 hops with a total RTT of less than 3ms. Then it bounces through level 3 (of all people) and once it hits the boundary router for NTT we're already at 60-67ms. The total routing time through NTT is sub 3ms. I don't have my home PC with the tracert I ran there for reference at the moment, but the majority of the additional routing time experienced was not within NTT. It's the routing of my ISP and their backbone peers.

This is what I have been saying the entire time, it's poor routing by the ISPs and backbone services for residential customers. However, due to that, and the number of hops involved, it's highly possible for that to be optimized with better routes and fewer hops. Looking at the tracert for my corporate net, the bounce through level 3 (starting in Atlanta) is what get's me to California and it's consuming almost 45 ms. Due to the hop to Atlanta (approximately 12-15ms) from my current location, my PC may as well be on the east coast distance wise.

The point of all of this is simple, optimized routing can make a *huge* difference to people's ping time. People with a 160ms ping from the continental US should be able to get under 100ms with better routing. Players in South America should be able to get improved pings too, again with improved routing. Taking those SA players who have gone from reasonable to ridiculous pings, I think you and I would have to agree that the initial hops that they take to the US are almost certainly the same now as they were before, it's the internal routing in the US that is causing the issue. And based on the traces I have seen, that's not an NTT issue it's a routing issue of the ISPs and backbone providers between NTT and the entry point from the undersea link. For example, for someone in SA seeing a 250ms increase in ping time, at least 200ms of that is coming from bad routing in the US. If that routing is optimized their ping should drop much closer to what it was before. But again, it depends on route optimization and of course network congestion.

You seem fixated on NTT as an issue, they are not. Their internal routing is not the issue based on the tracert outputs I have seen. It's the routing of traffic from the client to their boundary routers, and back to the client from that boundary that is the primary issue. That is something that can, and should change since the routing of the Internet is dynamic , it should optimize itself to a degree - depending on how many static routes are being forced in the path taken. Players complaining to the ISPs may help, but since NTT and SE are not the routing issue, complaining that SE should do something isn't going to help.

[Raikki]

You seem fixated on NTT as an issue, they are not. Their internal routing is not the issue based on the tracert outputs I have seen. It's the routing of traffic from the client to their boundary routers, and back to the client from that boundary that is the primary issue.

Except that for a lot of users, they actually are an issue. NTT's route from the northeast is simply not very good. There are much better routes into the area that are not available to us. All we can do is get to the closest NTT node the best we can, which in this case is NYC, and my route to them in NYC is already perfectly fine. I've tested my routes to multiple VPN endpoints and other datacenters in the area and get better routes to the majority of them.

[Namingaway]

Unfortunately that person doesn't quite know what they're talking about. It's possible that NTT could improve their internal routing in the future, but the location of the datacenters is set in stone now. The Colorado address you're seeing is a business address that NTT America's IP's are registered to, but is not where they're being used. This is abundantly clear if you simply look at their website:

http://www.us.ntt.com/en/services/da...center/us.html

They are not running a Colorado datacenter. They have facilities in California, Virginia, and Texas. No matter what conspiracies you want to entertain, even if you want to ignore the fact that the route terminates right after a clearly-labeled "sacramento, ca" hop, the fact is that the low pings northwest players are getting now would not even be possible if it were in Virginia or Texas no matter how good the routing was. You're being asked to believe the Earth might be flat.

Ah ok. That makes sense. It's nice to know that's why the geosites are saying that. I saw several Geosites putting it in Colorado when I punched in the IP address since I wasn't sure that scrmca was supposed to be brought out to be Sacramento due to the level 3 servers completely spelling out Montreal. I saw the tracert and now I realized that I am entering the NTT network in Palo Alto (plalca), then heading east to San Jose (snjsca) then north to Sacramento (scrmca). didn't think I would be going down there to enter the network. Thanks for the info as well! Now I'm able to see how it gets there.