Ehh I trust the reporting and generally agree that RTO was/is executed hamfisted but I dunno if this particular incident "makes" the narrative. IIRC LSE rate has been increasing for many years, maybe most of AWS's existence. This is part and parcel of building something so complex that continues to grow and evolve.
I do expect much better of them and they certainly have problems to solve but this is a big company evolution thing and not an Amazon-specific thing imo.
The real story from this incident is that Amazon’s “aws” partition doesn’t actually have multiple regions - effectively, it’s all IAD in a trench-coat.
This is a big deal. My employer has already started to look at bringing back our old racks from storage and switching back to on-premises. Cannot imagine he’s alone in that.
Can you elaborate what convinced you of this? We were running mostly in us-west and saw almost no impact, despite using a broad spectrum of AWS infrastructure and tooling.
I do expect much better of them and they certainly have problems to solve but this is a big company evolution thing and not an Amazon-specific thing imo.