I do Server Support for a major smartphone OEM and I can say that balancing Server load, redundancy, test environments, and integrity is a tricky thing. Bottom line, there is no way to truly test anything concerning load balancing until it is in a live environment. That is especially going to be true here. There is no way that BW is properly prepared to support a high end roll out. There's no way that any company could support that efficiently from all angles unless they had every server in the world running their software. This is why so many people cite the fact that MMO roll outs are usually the worst in regards to load balancing. IMO, they are using a good strategy here. It's obvious because nothing has actually failed in regards to up time yet. They see an issue, they take the Servers down, implement whatever fix comes out of the test environment, and then put it live again. The queues allow them to monitor heavy traffic statistics on a bulk level instead of running the risk of rebooting the servers, which means downtime anyway. It's being done better than I have seen most roll outs, and is being approached from a Network perspective rather than a run of the mill marketing scheme. With one, the other will suffer IMO.