My money is on it being related to a mix of the scourge of microservices and “cloud computing”, and a poorly planned and thought out implementation in their sharding tech.
In theory, the two are a perfect match for each other. In practice, well. This happens.
When a new shard is required, the “cloud” should be able to dynamically spawn a new “server” on demand. It keeps costs down and theoretical scalability high. It should be able to provide enough server resources to keep everyone on the shard running smoothly. If more people join this shard, more resources can be allocated. If people leave, those resources can be allocated elsewhere.
I’m guessing a poor implementation on multiple fronts leads to the issues experienced daily. Constantly loading and unloading these services, swapping players between them and possibly not letting them scale up as high as required.
It’s just a guess, but it’s not an uncommon conundrum and balancing act in the enterprise world. On-demand is popular with the bean counters as it potentially reduces costs. Only using resources when and where required, as opposed to outlying a set number of bare metal servers for each realm. It’s great when implemented correctly, and hell when it isn’t.
Of course with Blizzard presumably still running their own servers and not leasing out from cloud providers, it’s a bit more fuzzy, but it still works. Server resources can be dynamically shared between all their games. More people playing Overwatch compared to WoW? Give OW more of the server resources. They would not need to have a big box with an Overwatch label, a WoW label, a HotS label and a D3 label. They just have a bunch of boxes that are shared between everything as required.
Regardless of what the issue is, the fact it clearly wasn’t properly tested is poor form on Blizzards end. The fact they didn’t have a backup plan in case this technology didn’t work out is poor form. The fact they didn’t have a backup FOR the backup is poor form. You don’t rush head first with such major fundamental changes in architecture on a platform of millions of paying users without contingencies in place.
Blizzard doesn’t care. They’re too arrogant for “best practices”. Anything they crap out IS the best practice in their eyes. It’s seen through how they treat players with their gameplay decisions. It’s seen here. It’s seen in how they communicate. Blizzard is always right.
It might take so long to fix because it requires a fundamental rewrite of the technology behind lots of “behind the scenes” tech. It might take so long because they don’t know how to fix it given their existing architecture. It might take so long because they’re incompetent? Because Finance doesn’t want to spend the money to make what needs to happen, happen? Because they just don’t want to? Because they can’t?
No matter what the reason is - issues with scalability, issues with implementation, financial, technical or workforce related, it’s not acceptable going on half a year for a PAID service.
It’s bloody disgusting and not something that should be treated so casually from what would once be considered one of the, if not the premiere name in PC gaming.
Hey, at least when it’s all over, maybe they’ll give us a 10% discount on a store mount of our choice for the over half a year of “inconvenience”. Or more likely, they’ll “fix” it (with the help of some nice AH fix style “compromises”) and do nothing other than a half-arsed apology and thank their loyal customers for continuing to fund their inadequate, poorly performing game and the developers behind it.