3 ms·

Azure, Google Cloud, AWS and others need to have a “Status alliance” where they determine the status of each of their services by a quorum using all cloud provi

Azure, Google Cloud, AWS and others need to have a “Status alliance” where they determine the status of each of their services by a quorum using all cloud providers. Status pages are virtually useless these days

smt88 5y ago
They can do this without an alliance. They very intentionally choose not to do it. Every major company has moved away from having accurate status pages.
- dentemple 5y ago
  It's because none of these companies are held responsible for missing their actual SLAs, as opposed to their self-reported SLA compliance. So unless regulation gets implemented that says otherwise, there's zero incentive for any company to maintain an accurate status page.
  - iso1210 5y ago
    The only uptimes I'm concerned with are my own services that my own monitoring keeps on top of, this varies - if the monitoring page goes down for 10 seconds I'm not worried, if one leg of a smpte-2022-7 is down for a second that's fine, if it keeps going down for a second that's a concern, etc. If something I'm responsible goes down to the point that my stakeholders are complaining (which is something seriously wrong), they are not going to be happy with "oh the cloud was down, not my fault" If AWS is down or not it meaningless to me, if my service running on AWS is down or not is the key metric. If a service is down and I can't get into it, then chatter on things like outages mailing list, or HN, will let me know if it's yet another cloud failure, or if it's something that's affecting my machine only.
  - xtracto 5y ago
    >It's because none of these companies are held responsible for missing their actual SLAs, as opposed to their self-reported SLA compliance. Right, there should be an "alliance" of customers from different large providers (something like a Union but instead of workers, it would be customers). They are the ones that should measure SLAs and hold the provider accountable.
  - soheil 5y ago
    How did you find a way to bring regulations into this? There are monitoring services you can pay for to keep an eye on your SLAs and your vendors'. If not happy with the results switch.
    - ybloviator 5y ago
      Technically, there are already regulations. SLA lies are fraud. But I'm leery of any business who's so dishonest they fear any outside oversight that brings repercussions for said dishonesty. "If not happy, switch" is silly - it's not the customer's problem. And if you're a large customer and have invested heavily in getting staff trained on AWS, you can't just move.
      - soheil 5y ago
        A) don't build a business that relies solely on existence of another B) switch to another vendor if not happy with current vendor Really not that complicated.
  - cwkoss 5y ago
    I wonder if there could be profitable play where an organization monitors SLA compliance, and then produces a batch of lawsuits or class action suit on behalf of all of its members when the SLA is violated.
    - pm90 5y ago
      This is a neat idea. Install a simple agent in all customers' environments, select AWS dependencies, then monitor uptime over time. Aggregate across customers, and then go to AWS with this data.
- arch-ninja 5y ago
  Steam has a great status page, companies like that and cloudfare will eat Alphabet's lunch in the next 17-18 years.
  - ExtraE 5y ago
    That's a tight time frame a long way off. How'd you arrive at 17-18?
  - moolcool 5y ago
    Where do you anticipate Steam to compete with Alphabet?
- beamatronic 5y ago
  In a broader world sense, we live in the post-truth era.
LinuxBender 5y ago
Or just modify sites like DownDetector to show who is hosting each site. When{n} number of sites are down hosted on {x} one could draw a conclusion. It won't be as detailed as "xyz services failed" but rather the overall operational chain is broken. There could be a graph that shows 99% of sites hosted on amazon US East 1 down it would be hard to hide that. This could also paint a picture of what companies are not active-active-multi-cloud.
- avaika 5y ago
  Cloud providers are not just about web hosting. There are dozens of tools hidden from end users. E.g. services used for background operations / maintenance (e.g. aws codecommit / codebuild / or even aws web console like today). This kind of outages won't bring down your web site, but still might break your normal workflow and even cost you some money.
DarthNebo 5y ago
https://en.wikipedia.org/wiki/Mexican_standoff https://en.wikipedia.org/wiki/Mexican_standoff
- beamatronic 5y ago
  It's not the prisoner's dilemma?

smt88 5y ago

They can do this without an alliance. They very intentionally choose not to do it. Every major company has moved away from having accurate status pages.

dentemple 5y ago
It's because none of these companies are held responsible for missing their actual SLAs, as opposed to their self-reported SLA compliance. So unless regulation gets implemented that says otherwise, there's zero incentive for any company to maintain an accurate status page.
- iso1210 5y ago
  The only uptimes I'm concerned with are my own services that my own monitoring keeps on top of, this varies - if the monitoring page goes down for 10 seconds I'm not worried, if one leg of a smpte-2022-7 is down for a second that's fine, if it keeps going down for a second that's a concern, etc. If something I'm responsible goes down to the point that my stakeholders are complaining (which is something seriously wrong), they are not going to be happy with "oh the cloud was down, not my fault" If AWS is down or not it meaningless to me, if my service running on AWS is down or not is the key metric. If a service is down and I can't get into it, then chatter on things like outages mailing list, or HN, will let me know if it's yet another cloud failure, or if it's something that's affecting my machine only.
- xtracto 5y ago
  >It's because none of these companies are held responsible for missing their actual SLAs, as opposed to their self-reported SLA compliance. Right, there should be an "alliance" of customers from different large providers (something like a Union but instead of workers, it would be customers). They are the ones that should measure SLAs and hold the provider accountable.
- soheil 5y ago
  How did you find a way to bring regulations into this? There are monitoring services you can pay for to keep an eye on your SLAs and your vendors'. If not happy with the results switch.
  - ybloviator 5y ago
    Technically, there are already regulations. SLA lies are fraud. But I'm leery of any business who's so dishonest they fear any outside oversight that brings repercussions for said dishonesty. "If not happy, switch" is silly - it's not the customer's problem. And if you're a large customer and have invested heavily in getting staff trained on AWS, you can't just move.
    - soheil 5y ago
      A) don't build a business that relies solely on existence of another B) switch to another vendor if not happy with current vendor Really not that complicated.
- cwkoss 5y ago
  I wonder if there could be profitable play where an organization monitors SLA compliance, and then produces a batch of lawsuits or class action suit on behalf of all of its members when the SLA is violated.
  - pm90 5y ago
    This is a neat idea. Install a simple agent in all customers' environments, select AWS dependencies, then monitor uptime over time. Aggregate across customers, and then go to AWS with this data.
arch-ninja 5y ago
Steam has a great status page, companies like that and cloudfare will eat Alphabet's lunch in the next 17-18 years.
- ExtraE 5y ago
  That's a tight time frame a long way off. How'd you arrive at 17-18?
- moolcool 5y ago
  Where do you anticipate Steam to compete with Alphabet?
beamatronic 5y ago
In a broader world sense, we live in the post-truth era.

LinuxBender 5y ago

Or just modify sites like DownDetector to show who is hosting each site. When{n} number of sites are down hosted on {x} one could draw a conclusion. It won't be as detailed as "xyz services failed" but rather the overall operational chain is broken. There could be a graph that shows 99% of sites hosted on amazon US East 1 down it would be hard to hide that. This could also paint a picture of what companies are not active-active-multi-cloud.

avaika 5y ago
Cloud providers are not just about web hosting. There are dozens of tools hidden from end users. E.g. services used for background operations / maintenance (e.g. aws codecommit / codebuild / or even aws web console like today). This kind of outages won't bring down your web site, but still might break your normal workflow and even cost you some money.

DarthNebo 5y ago

https://en.wikipedia.org/wiki/Mexican_standoff https://en.wikipedia.org/wiki/Mexican_standoff

beamatronic 5y ago
It's not the prisoner's dilemma?