So you have information related to their back end we don’t? Because that’s very clearly not the case. If you shard people together you need to share state already. The backend databases (Oracle at one point) can easily handle that throughput. Given that it’s a shard by shard thing the atomicity is per shard not per connected realm. E.g. even if you and I are on the same realm but in different shards there is nothing I can do that will affect you. Thus from a database perspective we’re completely independent.
Moreover high throughput systems only commit results to databases, the actual current state is stored in memory.
not to mention you take the US Normal Servers in the Chicago datacentre for example, well 47 of the current 101 servers are located there, will be 44/95 if all connections go through but thats still alot of servers and basicially double what any other region except maybe china would have, but even in China they have 4 datacenters i think.
As Far as data centers in the US servers goes:
Chicago hosts 47 Normal US servers, 5 US RP servers, 4 Portuguese servers and the 3 spanish servers
Los Angeles hosts 29 US normal Servers and 6 US RP servers
and obviously Sydney hosts the now 7 Oceanic servers.
Sure thats the goal, but this doesnt necessarily mean faction balance because that would be impossible to achieve, it fundamentally doesnt exist and players are fluid and can move.
There is no reason for this, the amount of servers doesnt really matter and there arent that many in OCE anyway, if 1 server cluster is 6 servers and another 1 and they come out to the same population thats what matters.
Like i said, doesnt exist, Blizzard cant be constrained to it.
They have proven many times throughout the connection programme that they are willing to merger former PVP realms with former PVE realms.
Not correct at all. If we are on the same server and have brutosaurs, and you need a flask that’s not on the auction house, I can list it from my shard, and you can buy it from your shard.
Meanwhile, people on your shard from different servers cannot see my listing, nor can you see theirs; they also can’t trade the flask you need to you.
Everything that’s persistent is stored with your character’s server’s database. Shared state for shards is limited to ephemeral information that, for example, would most likely go away in a server crash.
If that brings the server to a halt… then they have bigger architectural issues. Markets run every day with Billions of transactions per second. The AH isn’t that much of a load and can be easily partitioned.
Don’t assume it’s a single database, or more importantly don’t assume it’s not already a partition(s) on a single database. Oracle is insanely expensive. So while they are definitely running replication running multiple extraneous instances wouldn’t be cost effective given the highway robbery that is oracle licensing.
But honestly I’m not going to respond further, I work daily with systems that have larger volume. It’s not an unsolvable problem. As a software engineer sometimes you need to get creative, I know I have many times.
I’m wondering the same thing. It looks like a bunch of the other servers being connected today are offline, but we’re still online. Maybe our merge is being delayed?
Edit: I just got confirmation from Blizzard customer support on Twitter that this group is not getting merged today, and will be rescheduled.