Prompt
Let me guess. Prompt was missing the "Ensure this time it works." phrase at the end.
Amazon Web Services’ US-EAST-1 region, which last week caused massive disruption to online services, is having another bad day as internal dependencies again prove problematic. At 3:36 PM PDT on October 28 (10:36PM UTC), the cloud colossus advised customers that “Earlier today some EC2 launches within the use1-az2 Availability …
"Ensure this time it works." phrase at the end.
Really have arrived back at ritual magic and incantations to summon and importune daemons to perform various foolish tasks.
Perhaps AWS should employ the shade of Aleister Crowley to tame their diabolical legions although a good old fashioned exorcism followed by incineration of the ranks of heretical manglement as an act of faith might be more satisfactory or at the very least satisfying.
Your comment was prophetic: Heathrow, NatWest and Minecraft sites down in Microsoft global outage (4pm GMT 29 Oct).
"The Cloud" is just somebody else's computer that you pay an outrageous set of fees to use.
But "The Cloud" computer is managed by human beings in the end, and controlled by largely human-written software (which typically puts so called "vibe" code to shame for quality and reliability in reality), and mistakes will happen.
But when "The Cloud" makes a mistake, it makes it for thousands and tens of thousands of systems. Not just one corporate server cluster.
After seeing my bill on trial runs of cloud services, I rapidly realized I could buy self-hosting hardware for about 4-5 months of cloud services if I had a real workload. I didn't buy self-hosting hardware, but it convinced me that self-hosting or co-location typically put "The Cloud" to shame for price, even when you factored in some decent number of sysadmins (you need sysadmins for the hosts you run, not the Amazon EC service configurations, but that's "six of one; half a dozen of the other.")
"The end is near for the luddites"
Depending on your time scale, the end is near for everyone. "Nothing lasts forever but the Earth and Sky" (Kerry Livgren Dust in the Wind 1977).
BTW, the luddites were pretty much right. Their relatively comfortable cottage textile weaving industry was replaced by a more efficient, but by modern standards quite gruesome early 19th century industrial complex. Something somewhat similar may be happening today. Except that it's quite unclear that much of value will emerge from the chaotic, wildly insecure, dubiously reliable, shambles that is being created.
This time it seems to me disturbingly likely that the luddites will "win". At least in the sense of losing less than their opponents.
A sprawling, complex cybernetic entity with countless interdependent limbs and organs representing its myriad services and components.
Where a single misfiring DNS neuron or corrupted synapse triggers a cascading apocalypse of failures across its tangled neural subnet. Like a digital hydra severing its own heads only to spawn a dozen more glitches. This creature breathes in API calls and exhales erratic error codes. Its sprawling, fractal limbs endlessly spawning clusters yet riddled with internal parasitic dependencies that twist resilience into fragility.
It sprawls and mutates in a dark silicon swamp, endlessly attempting self-healing rituals while cracking under its own overly complex and maddeningly recursive wiring. A deranged AI’s monstrous experiment in cloud life that’s both breathtaking in scale and horrifyingly unstable. Doomed to twitch and falter in cycles of brilliant creation and catastrophic collapse.
What you are describing rather resembles organisms called slime molds (https://en.wikipedia.org/wiki/Slime_mold). Slime molds consist of a rather diverse selection of single celled organisms that will, in the right circumstances, get together with their mates and organize into a sort of pseudo-organism capable of doing things the individual critters can't do.
Slime molds are weird.
So, if you ask me, is today's internet.
> This creature breathes in API calls and exhales erratic error codes.
I missed a connecting flight last week. I was sent to the transfer desk to get rebooked. The transfer desk is now a bank of self service machines with human slaves. On entering my details, the machine returned, believe it or not, “Generic Error”. WTF? A senior human came and took me to another desk where a lady sorted this out in less than five minutes. How much did these machines cost?
And … at the same airport the new passport e-gates might scan your passport but are staggeringly missing the UI to tell you what to do. It currently displays only numbers, not words or pictograms. This is not a new feature, e-gates have been with us for some time. Perhaps the HMI was an optional extra …..
"Which of the two costs has the larger effect on Amazon's profits ?"
That may depend on how deep the pockets of the victims are. Some are "very big names" and the previous AWS DNS-related outage and today’s Azure DNS-based outage have potentially cost some of those big name a lot of money. Weasel words in contracts won't stop the sue-balls flying.
Or on the back of last weeks AWS snafu, have there been all sorts of niggly issues all over the internet.
I am currently "working" (as in I am doing their jobs for them) with a robot vacuum cleaner manufacturer who appear totally stumped as to why their system has stopped working with Google Home.
And the fact the symptom is a 404 error in the OAuth process really does trip my spidey senses.
The thing is THEY DON'T KNOW if they are dependent on AWS.
From what I saw in the office today, it only seemed to be affecting new log-ins. While it was going on, there were users happily not noticing anything wrong while anyone trying to authenticate or start a Teams meeting where some participants had to log in were having issues. It was close to end-of-day, so most didn't seem too bothered :-)
Back in the day, we convinced the beancounters to pay for UPS's because we found out if the power tripped, we'd get in a mess.
When we bought a new server, we kept the old one on a shelf - because it could be pressed back into service if one of the newer ones fell over. Beancounters generally approved.
Now we put everything in the cloud and, for some reason, the bean-people can't be convinced that we really need to have a fall-back because we're now entirely reliant on a dystopian collection of software and services, all provided by external businesses we have absolutely no control over.
Are we genuinely regressing here or simply unable to learn from the past?
"AWS drives companies of all sizes to be industry game changers with cloud technology that fuels innovation.