I have to deal with a number of folks who will be overjoyed to read this news when their tech cartel vendor of choice forwards it this evening.
There's a huge contingent of currently endangered infrastructure folks (and vendors who feed off them) out there who throw a party every time AWS has a visible outage.
AWS sucking at availability (and especially specific parts like EBS, and then services built on top of EBS and on top of AWS) doesn't mean the correct option is to mine your own sand and dig up your own coal to run servers in your own datacenters in the basement.
Even if you're totally sold on the cloud, you can still have a requirement that things be transparent all the way down. AWS is one of the least transparent hosting options around.
If you're a customer of a regular colo, or even a managed hosting provider themselves based at a colo, it's pretty easy to dig into how the infrastructure is set up, identify areas where you need your own redundancy, etc. Essentially impossible within AWS -- there is no reason intelligible to me that ELB in multiple AZs should depend on EBS in a single AZ, but that's how they have it set up.
I'd question if they are truly wrong to send out news like this? Basically you have to weigh the realities of the 'cloud' with the perception.
The perception that many people has is that somehow the 'cloud' is a magical up-time device that will save you money in droves.
The reality is that for many companies with mid-level traffic and aren't a start-up with a billion users, the 'old' style tech might very well be your best option.
I have to deal with a number of folks who will be overjoyed to read this news when their tech cartel vendor of choice forwards it this evening.
There's a huge contingent of currently endangered infrastructure folks (and vendors who feed off them) out there who throw a party every time AWS has a visible outage.