So some diligent work by some of the other cats paying close attention to the connection process has revealed an interesting detail, but first, a bit of set-up.
The current state of WoW’s realm structure going into this process involved four datacenters where the realm hardware is housed, which are located in Los Angeles, New York, Chicago, and Phoenix. Realms in the same datacenter are the pool of realms that players are usually drawn from for all the cross-realm shenanigans like CRZ, LFD, Battlegrounds, and LFR, but of course if you’re doing premade cross-realm stuff like BattleTag/RealID groups, it doesn’t matter what datacenter your realm is in. Also, all of the datacenters have realms that are in all different time zones, so the location of the datacenter isn’t an indicator of what time zones or regions are served there.
1. Realms in Phoenix are only being connected with realms in LA, while realms in New York are only being connected with realms in Chicago. (There are some connections that are exclusively between LA realms, and some exclusively between Chicago realms, but that’s not the case for Phoenix or New York.)
2. Consequently, once a connection is made in, say, a Phoenix realm, players who log into that realm have their traffic directed to a new address in the LA datacenter. Same goes for New York realms, with their traffic going to Chicago.
3. As a result, there are some players reporting increased ping times to access their realms, but since there hasn’t been a really dramatic outcry about it on the forums, it doesn’t appear to be something most players have noticed.
The major thing this revelation does is help to narrow down the possibilities for which realms get connected to which, especially once you get into discussing the RP/RP-PVP realms or the non-North American regions served by these datacenters. And while it doesn’t create a universal rule (for example, the Boulderfist/Bloodscalp/Dunemaul/Stonemaul/Maiev CR draws four realms from the same battlegroup, and only draws one realm from the opposing datacenter, and some other connections don’t draw from the opposing datacenter at all) it does create something that can help predict how the rest of this is going to shake out.
It also explains one little mystery: Anvilmar and Undermine were originally announced to get connected to each other back in Round 7. That was put on hold before Round 7 was implemented, and later on, the pairing disappeared completely; not pushed to a subsequent round, but dropped entirely from the list. After comparing all of the existing and proposed connections together, it turns out that Anvilmar/Undermine was the only connection that drew two realms from New York, without connecting first to one in Chicago.
I think it’s still likely that those two realms will ultimately be connected together, but they will likely get connected to a Chicago-side realm first.
So, that being said, what are the rules that seem to govern connections going forward?
- Realms in PHX will be connected to realms in LA, and realms in NY will be connected to realms in CHI. Connections may happen in CHI or LA that don’t involve realms in NY or PHX, respectively, but the inverse is not true.
- Realms will always be connected along server types (PVE/PVP/RP/RP-PVP).
- Only one realm is added to a connection at a time.
- Realms will always be connected along region/language barriers. (There’s no evidence to support this aside from the fact that no Oceanic, Brazilian or Latin American realms have been connected yet.)
As for targets, it seems that the existing CRs are topping off between 5-6000 raid-capable characters across both factions. From this, we can infer the following:
- Realms that have more characters than the existing CRs will likely not be connected.
- Realms that can’t be connected with another eligible realm without going over the limit (and thus threatening to become a high-pop queued realm) will likely not be connected.
- Realms that cannot be connected because no eligible realm is available may be left unconnected.
If we look particularly at the three non-North American zones serviced by the NA datacenters, we can actually draw some conclusions:
1. EDITED: Out of the three Latin American realms (Ragnaros, Drakkari, and Quel’thalas), the former two are PVP, while Quel’thalas is PVE. realms. However, only Drakkari is under the 5-6k threshold, at 3.7k characters. While that’s not an unhealthy population it could certainly be better, but there’s nowhere to go for Drakkari, unless Blizzard opens up some FCMs from Ragnaros. (Thanks to the anon commenter Me for pointing this out.)
2. There are five realms for Brazil, with two PVP and three PVE. Azralon, Goldrinn, and Nemesis are all above the threshold, while Tol Barad (PVP) and Gallywix (PVE) are well below the threshold, and might potentially hit it if they were combined. However, they’re opposing server types, meaning they can’t be combined… which is distressing, given that those are the two worst-progressed Brazilian servers, and are in the middle of many of the other realms that are being connected currently. As with the Latin American realms, FCMs could fix this, but it clearly not a priority.
3. There are twelve Oceanic realms, six PVE and six PVP. Almost all of them are in the 3-4k character range, meaning they can’t be combined with each other without going over the threshold. Out of the whole set, the only viable connections possible involve Gundrak and Dreadmaul, both PVP servers with relatively low populations and middling progression. While they could be combined with each other to possibly compete with the rest of the Oceanic set, they could also be used individually to buff the next two smallest realms (Thaurissan and Jubei’thos). The other two PVP realms are Barthilas, which is already over the threshold, and Frostmourne, which is the #3 server in the NA and DOUBLES the threshold as a super-high-pop realm.
As for the RP/RP-PVP realms and the remaining North American realms, that’ll have to wait for next time.