Seeing all these horror stories, I can't help but wonder when people are going to switch back to on prem, or at least some hybrid. It's pretty apparent we can't trust cloud providers...
I think it's a vocal minority. The vast majority of GCS customers won't have this problem. These kind of posts typically do very well on HN.
It doesn't take most to have this problem. It takes a reasonable proportion to fear they could have this problem.
That said, I think cost is more likely to drive change here than fear over billing - most people pick lover cost over risk mitigation far more often than they'd like to admit.
The same is true of DigitalOcean, fwiw. Shutting down machines used in production with no notice, outside working hours. I don’t know why these companies don’t care - they’re massively shooting themselves in their feet. Incidents like this turn customers off for life.
> outside working hours Working hours are relative. Our support routinely gets awoken at like 3am because someone in Singapore, where one of our larger customers has outsourced their IT to, just started their day and decided to follow up on a case. Though I'll agree that weekends should be considered off-limits regardless.
> Working hours are relative
Sorry but for me this doesn't cut it when you have staff and availability zones in every continent on the planet. It isn't some mom and pop shop in one town. It shouldn't be much to ask a company of Google's size and breadth to consider your local timezone for urgent communications.
The point is not to limit the size of the industry. It is to limit the size of a company . Think of a oil refinery as a monolith, and that such a mandate would enforce that you'd have to build the whole system a set of independent microservices. The industry as a whole would still employ a large number of people, but now they would be forced to coordinate through specific interfaces (the smaller business units) inste…
Capitalism would not allow such a company to exists. There would be too much bottlenecks and the left hand not knowing what the right hand is doing scenario and the end product would be more expensive compared to a company with no such limitation. Buyers would prefer buying the product at much cheaper rates and then just transporting it to their locations.
I could easily argue the opposite: in a scenario where all companies are reduced in size, they would be forced to become a lot more efficient in their communication and they would be a lot more open and agile.
Also, a lot of the work done by people in Big Corp is something that is not outsourced merely because they don't want to let their competitors to have access it. A lot of these redundancies would be eliminated and replaced by smaller units that can serve multiple companies. Companies that have internal "business systems" developers would simply get rid of that and looking into third-party SaaS and/or open source solutions that could be used, etc.
If you are paying for the load/pressure, what's the problem?
Backblaze is very cheap and has several "don't overload our system" clauses in their terms and conditions. They're cheap and reliable, that generally comes at the risk of early termination. I don't think you can have a cheap, reliable, unlimited service; pick two at most.
They probably meant their B2 storage service, which is like S3 and bills for storage, egress, some operations, etc. That’s one service I wouldn’t expect to have some cancellation for overload in place. You’re using a token anyway, why not just rate-limit you instead?
Right, but I think your sarcasm was misplaced. It is actually possible for customers to make a mistake. These mistakes are relatively easy to prevent. It could be that in this particular case there wasn't anything the OP could have done and they are simply unlucky. But that's very unlikely.
> These mistakes are relatively easy to prevent. Assuming one knows about them. How would one learn about them, without making the same mistakes themselves?
Most of it will actually be documented by your provider, and is often pretty obvious. For example, if you're paying with card, have at least one extra as backup payment method (which most providers allow and recommend). For extra safety: make one of these cards a pre-paid debit card loaded with enough money to pay for at least one month. The one thing you don't want to happen is to miss a payment, which could happen for many reasons that aren't under your full control, like a technical problem with one of the cards, a fraud detection false positive rendering your card invalid, etc. And once you are unable to complete a payment, that's when the trouble begins.
Right, but I think your sarcasm was misplaced. It is actually possible for customers to make a mistake. These mistakes are relatively easy to prevent. It could be that in this particular case there wasn't anything the OP could have done and they are simply unlucky. But that's very unlikely.
All - absolutely all - the evidence we have seems to show that Google was completely wrong - that a payment didn't go off because of a known Google outage, that the company then made a manual payment, but started to receive threatening messages, that they appealed to billing who told them they were safe, but then the site was cut off in the middle of the night on the weekend. Your argument appears to be this: "Facts…
Epistimological check-in: we don't actually have any evidence, except for a very high level story from the OP without much relevant detail. So all we have is the base rate, which is in almost all cases when there are billing problems they are at least to some extent because of negligence or fraud on the part of the customer.
> Anything less than that and all you are getting is an incentive for the make a risk calculation between profit and losses of being caught. But that logic applies to anything that any company does. You want to effectively kill any company for any infraction? That sounds a little too severe.
If the infraction was intentional and with the sole purpose of getting unfair competitive advantage in the market, yes, the company should be dissolved for it.
> If the infraction was intentional and with the sole purpose of getting unfair competitive advantage in the market, yes, the company should be dissolved for it.
My memory may be hazy[1], but I don't think Dieselgate was ever proven intentional; your proposed policy wouldn't have kicked into effect anyway.
[1] So feel free to post links to the findings if you have them handy
It sucks that they gave you conflicting information that resulted in this but I wouldn't trust having my cloud infrastructure payments be handled using a single credit card. If you have a spend of over $2500/mo you can apply for invoiced billing directly from Google or if you have less than that you can contact a GCP reseller partner to get invoiced by them. Some GCP reseller partners also have a close relationship w…
$2500 is honestly not that much. Google cloud dropped the ball on this and shouldn’t need to be handled with special relationships
What is much and what is not is relative. $2500 breaks the bank for tiny startup and is a rounding error for Google.