I'm struggling to be honest; on the one side Blizzard has 20+ years of experience with large scale launches and nowadays access to all the servers and compute they could need as long as they pull out a credit card (this goes for all large scale projects really). This has been a thing for... actually AWS is coming up on having been a thing for 20 years now. In theory, scaling and day 1 launches and peak load have been a problem that has been looked at and solved for 20 years now.

I don't, therefore, really understand why AI would have made such a big difference this time compared to the launch of Classic. But I can only dream of projects and launches of such a scale.

▲

rowanG077 a day ago | parent [-]

I'm perplexed why you just apply the learnings that a company like AWS has gotten to a company that has up until recently clearly been struggling with scale.

That some company has solved something doesn't mean another company can just use it. You can't just "from aws import scale" at the top of your fresh new project and be fine with millions of concurrent users.

	▲	chrisfosterelli a day ago \| parent [-]
		I think you maybe misunderstood the parent comment, I don't they were saying that AWS figured out scaling their physical infra so therefore Blizzard could figure out scaling their physical infra. I think they were saying that tools like AWS -- which provide access to near infinitely scalable infra on demand -- are now commonplace which make scaling greenfield projects much easier than it was 20 years ago. I don't know about blizzard but a lot of game providers do run on hyperscalers nowadays so I understand what they're getting at.