Google's problem has always been customer support. Not just the process but the idea that your customers deserve to be supported when using your product.
You shouldn't ask people to migrate before having a documented, tested, tool-assisted, clear and intuitive migration flow that takes very little time.
I don't know about the 5 minutes, but the CTO replied when I complained in some comment and asked me for a trace. That was extraordinary user service (I am not even a customer), but the problem has not really been solved some 3 months later after sending the trace. One error has been replaced by another one...
I'm also completely sick of fielding API migration emails from Google on mobile Android/Firebase. Feel like I get some "action required" every couple of weeks. This stuff saps our resources, both to fix it, and to diagnose if the issue even applies to us in the first place. If you are lucky Google includes details of the app using the API, but often this part is even left out.
Yeah the amount of disrespect of developer time they show here is crazy. You could have made the billing api backwards compatible for 10 years, its a super small api, but instead they force breaking changes every couple of years
Maybe it is a strategy for cleaning up old apps or something, but I doubt it
I would say that Googlers are painfully aware of this. But the real problem is leadership. It starts with the CEO and various execs. Their employees are still living in the "growth hacking" and highly academic world (at least their engineers) and their execs are drunken with arrogance, convinced their Titanic will never sink. And they may be partially right, it may not sink but it will never regain its former glory. It will rot on the shelf, besides IBM,AT&T, HP and many more.
At its core, alphabet is a services company, it does not sell goods (to the most part). Their board needs to decide if the company should focus on business services or consumer services and replace the executives with people who have experience in that area.
The actual quality of GCP is superb in my opinion. They have a flair for architecting excellent solutions. I prefer GCP over any cloud from a purely technical perspective. Their leaders just don't get that it isn't enough. They're not operating a toy or a museum of technical marvels. Actual people need to use it.
I worked on GCP, left Google about two years ago. Some Googlers are aware of this, fewer care, and fewer still act on it - you sort of have to be an insane employee* to do that.
I saw tickets from high-paying, top-10 customers, go unanswered for days; no one senior enough on the team to answer felt it was more important than the day-to-day, and no one from the support/account executive staff felt they had the authority to demand it.
I see it otherwise. I think solving customer support is crucial to GCP's success, and since I agree with you that GCP is the better underlying product, there is a deluge of money passing Google by, just waiting for the right executive team to start caring about this to pick it up. Kind of like Microsoft pre-Satya.
This is really sad to read. I agree with you that GCP has the superior product. Azure is a fucking joke. AWS sucks but they do respond to feedback and will even get engineers on a call with you if you hit certain edge cases. The level of apathy you describe will very quickly sour users.
After leaving Google I started working at a startup that runs its production environment on AWS. We debugged an issue and experienced exactly what you're describing: our MSK cluster refused connections. Eventually that was solved by rebooting something on their side. It took less than 24 hours from when we opened a ticket until the issue was resolved, with a few back-and-forth messages between us and the AWS support engineer.
I spend around $20 a month on my personal AWS account, and the level of support I've received on the handful of occasions I've needed it has been phenomenal.
I'n guessing that's because AWS support engineers live in eternal fear of ending up on a PIP if I rate them less than 5 stars.
They're a trash company, and anything you get from them, even when you pay for the storage, is at best accidentally delivered to you and they could roll over in their sleep at any moment and snuff it all out.
Even grosser is that they explicitly limited the iOS version of Google Photos to refuse to work with scoped access. So on iOS the app will refuse to work without being granted whole library access, a permission that Google no longer allows to any third party photo app on Android. Blatantly anticompetitive.
A couple of months I was evaluating Dialogueflow for a chatgpt like ReAct agent I was working on. They have this blue "contact sales" button where you can chat with the sales team. Easy enough I thought, jumping right into it to ask my questions. I am greeted by a virtual guide that doesn't help. Then I ask for a human "Live chat with sales". Again another bot. How did I find out? The bot didn't bat an eye when I informed them I have 6 Billion customers on Mars, Venus, and Pluto. I ended up using streamlit.
This is why everything Google feels evolved, nothing lasts long enough to be obvious legacy bullshit.
Cut backs have finally forced AWS to sunset services, but they just hard refused for many, many years. People like using it because the workflows evolve, but AWS will not force migrations on you, sometimes to your own detriment.
When you first build an app, everything is fresh because developers haven't had to hamfistedly shove assumption breaking models and patterns into their program yet.
Cloud customers want to evolve to the latest tech, but redoing 20 years of the company for no business reason makes programmers happy but executives nervous.
AWS used to just bake keeping the light on into the cost of launching a product. They wouldn’t launch something they weren’t ready to support until the sun becomes a supernova.
Sadly, that seems to not be the case anymore, and it’s a worrying development. AWS giving up on their customer obsession makes it easier for the other CSPs, who aren’t as customer friendly, to compete.
For one, the OpsWorks configuration management tool is being replaced with Management Console, which is more encompassing, but different. You have to manually migrate or rebuild any deployed resources from one to the other.
From the outside I dont perceive the promotion dynamics for engineers working on old services is that different at AWS than Google or Azure. Finance and engineering do conspire to kill off things despite customer annoyance, perhaps as they should.
There is also a constant churn of retiring EC2 instance types, security certs, old database versions, EMR versions, OS versions, security practices, cost optimizations, etc to suck up your time with minimal business value. This is not different fundamentally than on prem, you just have slightly less control and an outside party forcing your hand to do the right thing isnt purely a bad thing...
AWS is pretty good about giving you a grace period to migrate off and informal warnings if they strategically want you to move to a different service of theirs before stuff gets killed if you are looped in with a TAM (e.g Data Pipelines vs more-expensive Glue). They seem to have recently migrated to a strategy where they disable services for new customers but actually dont kill them completely off but keep them on life support.
The only product I know well in your list is pre-VPC ec2 instances, which (to be fair) are a terrible product : it is cross tenant (as in : you can impact other customers). Good riddance.
I think many AWS services are well designed : isolated software-only components. They build a lot on top of a very stable infrastructure (VPC / s3 / ec2 / IAM), which means supporting a service is really cheap : they are just a couple of containers running somewhere
I completely agree. The CLI tools are fantastic, and the logging tools are much more user-friendly than AWS. Plus, the console manager is way better than Azure’s clutter. But the decision about domains was a huge misstep. It used to be so easy to buy a domain from Google and get a server running quickly. Why make the user experience so frustrating right from the start?
They definitely lost my trust with Google Domains, somehow I thought that such a foundational infrastructure layer will not be axed as easily, oh boy was I wrong..
A company that maintains that many TLDs is expected to have long term commitment to the domain related businesses but it turns out to be just a fleeting interest.
What cloud storage solution do you use instead? That's been one of my problems for things like photos and documents so I've just stuck with Google One.
I signed up when I had a Pixel and now I'm back to iPhone so I'm subbed to iCloud now, also because it includes HomeKit cameras recording functionality.
I used Google Photos on the iPhone for a while though, it worked fine you just have to open their app so it syncs.
Google Domains shutdown was an own-goal. They should have just ran it at a loss forever if that is what it took, the optics were and are just awful.
I get it, it was a "consumer" product essentially, hence selling the business to Squarespace instead of someone like Cloudflare. But anything related to DNS is going to make infra folk very wary of what else you might be willing to kill or neuter. It's just at that level of fundamental things that make operators skittish.
Google Domains was a service where no new features could be added, so nobody would get promoted for working on it, so nobody wanted to maintain it. Classic Google.
> I get it, it was a "consumer" product essentially, hence selling the
> business to Squarespace instead of someone like Cloudflare.
Ironically, I moved all my Google Domains domains to Cloudflare. Their revenue from being a domain registrar is likely a rounding error compared to their other products, but (1) now they have my credit card on file for value-add services, and (2) sometimes people with corporate spending authority ask me for advice about who they should buy cloud services from.
Grocery stores don't make their money from selling bread and milk, but a store that doesn't sell bread and milk is run by fools.
They shouldn't have any revenue from registering domains because they sell them at cost. It's probably the opposite because of all the overhead (which is of course compensated by pulling in more clients as you explained).
The obvious go-to choice was Cloudflare for us too, but then it turned out you can't use CF just as a registrar (at least on the basic plan or equivalent), you need to use Cloudflare's nameservers as well... But we use Google's Cloud DNS for everything, so that was a showstopper. In the end we went with AWS Route 53.
CF could probably get a lot more customers if they would allow you to use custom nameservers for your domain.
I don’t think it’s the case even on a paid tier, if it is I can’t find any reference to it. There are a lot of posts on their community support form where the recommendation is to use another registrar. https://community.cloudflare.com/t/use-cloudflare-registrar-...
That option means "your nameservers are now ns1/ns2 on your domain but they still map to Cloudflare", not "you get to pick your own third-party nameservers".
> CF could probably get a lot more customers if they would allow you to use custom nameservers for your domain.
Why would they want that? The whole point of CF as a registrar was so you could use the other services. The registrar is sold at cost. It's a way to lock you in.
It would lower the barrier to switch to Cloudflare for new customers, and once they are inside with a credit card on file it’s arguably easier to explore their product offerings
The funny thing is it wasn't even a true shutdown - Cloud Domains continues to be a reasonable registrar with proper IAM management so actually much nicer than Google Domains for larger projects. There were plenty of docs on migrating which was a one command thing (don't think there was a UI). Non-"consumers" had a reasonable out, which they could then drop for another provider if they felt a need, but there was definitely no requirement to end up at Squarespace.
But they never marketed it, among all the backlash over turning down Google Domains there was never any true call-out to this. To this day I suspect many people don't know about Cloud Domains (which still exists and accepts new domains). I can't fathom any user-related reason for this - the contract in selling to Squarespace must have forced Google to be unable to properly market Cloud Domains as a transition option. If not, the Google Domains PM truly didn't know the existence of Cloud Domains?
I was also surprised Google didn’t really mention the Cloud Domains path. I did it, but there are some caveats. It doesn’t support all domains that Google Domains did (e.g. “premium” domains on new tlds), and Squarespace does still end up being the registrar of record; Google Cloud is just a frontend.
Squarespace's domain panel crashed with a nondescript error when I tried to update nameservers prior to transferring out, and they shut off the Google nameservers as soon as the transfer went through on their side. To add insult to injury, Squarespace makes you wait 5 days for a transfer, with no way to expedite -- and in my case, they waited 6 days, taking me offline on a Friday night. This was the worst experience I've ever had using a domain service.
I believe the reason for divesting from the registrar business had to do more with ICANN rules that require Google not both be a registry operator and a registrar. Google does have a large registry business with 60+ TLDs so they could not also be a registrar (even if not for the same TLDs). It wasn’t really a money thing AFAIK.
I just rented servers on OVH instead. I mostly do CPU intensive and highly parallel computations, and went from 3500€ to 300€ for my usage.
As a startup I often put simplicity and ease of use before pricing but in this case that was too much of a difference!
Same. Cloud Run is a really nice product. It makes deploying containers a breeze and simpler than AWS's equivalent service (ECS / Fargate.)
I also found GCP's developer tools and docs to be quite solid, better than AWS. Example: It's nice that they provide Terraform snippets for most of their resources.
It's the least worst solution for "whole-workplace" systems, but most people who are saying "won’t use Google for anything cloud" probably self-host or use something like PurelyMail (https://purelymail.com/) or FastMail (https://www.fastmail.com/) for their own emails
Office365 is way better. You get native apps, which are obviously better than a web app, you get device integrations for enterprise, you get more advanced tooling, more features. Plus MS has better compliance options. If you're actually running a business, and not just a startup employee who doesn't need to do much, MS is the way.
Not sure why you got downvoted to dead. In big companies, M365 is a way more common and complete solution compared to anything else out there. This is just a fact - I actually really like e.g. Google Sheets, but the reality is that M365 can integrate with your Microsoft stack top-to-bottom and has low-code app, automation and database services that allow you to build out a good chunk of common LOB apps with minimal complexity (not that Dataverse doesn't have some ridiculous issues).
In my corner of EU, Office365 is the only thing I see everywhere(also in some government offices). I have yet to see/use GSuite anywhere yet. Also, M$-Azure is common on more ancient businesses, modern startups sometimes have AWS and hip startup in rare cases has GCP(when product does not involve personal data).
Surprising - as with the earlier commenter in EU, here in South Africa, including the multinationals spanning the continent, everything is Teams, including the government.
Sorry I’m the one who thought Google Workplace was the best, I’m learning something.
- Does Chrome keep nagging you for login and bookmark sync, when you use O365?
- Google’s way of handling shared inboxes is awful (Want an extra email? Create a …Google Group! Then assign members, tune the perms, etc.) How does it work in O365? Can I just create an email and assign 7 users on it?
I think it's a weird quirk of HN - we're encouraged to downvote, but I believe you can kill a comment if it gets just a few downvotes initially, which I think is overkill?
Anyway: big corps disable Chrome sync; most people will use Edge in enterprise because it integrates seamlessly into M365 and supports multiple work profiles (e.g. privileged accounts, service accounts, etc.)
Shared mailboxes are built into Active Directory; most enterprises have automated ways to create those -- usually the big concern is managing removing members when they're not supposed to be members anymore, hence the centralized management. There are also M365 groups you can create yourself, but that's just a distribution list. And generally, Teams is the way to go for any kind of collaboration work now - shared mailboxes seem to be a bit of a relic, usually meant for external interfaces that require an email address.
Matthew 7:27 “And everyone who hears these words of mine and does not do them will be like a foolish man who built his house on Google’s ever-shifting infrastructure.”
It’s true. Two weeks ago, I did a bit of light consulting for a data engineering project involving an IoT device and half way through designing the GCP architecture of the pipeline, I realized that Google was shutting down their IoT service… I ended up having to recommend AWS.
Google lost the rest of my trust with the ongoing fiasco with google play dev verification where many developers are not able to confirm their phone numbers for months and are at risk of losing their accounts. I can totally see myself being banned from GC for some nebulous reason or being locked out of account with no recourse. Their support bots are useless
That worries me in general about Google. You will trip some wire in some garbage code somewhere and that’s it, you’re locked out and banned for life with effectively no recourse and no hope.
Google domains was not integrated into google cloud.
And artifact registry is the better product and you had like 3 years time to migrate , which was a simple button press and changing the urls of the images. Of course it took some work, but it is not more expensive, effectively it’s cheaper since you can more easily clean up old images
The URLs might be mentioned in many different places. Why break URLs just to make a "better product"? Why couldn't the new product keep old URLs working?
Their end-users are advertisizers spending millions. For them Google is probably very responsive. The rest of us are users of pet projects at best and product at worst
The term end-user does not refer to advertisers by definition. Typically most people use the term "user" to refer to humans.
You're not wrong, but advertisers are a cancer on society that not only do not contribute any value but actively destroy the world around us. it's difficult to assign anything but deeply negative value to their needs and concerns.
"More than half a million UniSuper fund members went a week with no access to their superannuation accounts after a “one-of-a-kind” Google Cloud “misconfiguration” led to the financial services provider’s private cloud account being deleted, Google and UniSuper have revealed."
I think the latest public information was that it was a VMware "private cloud", a set of VMs, in GCP which accidentally had an expiration date set. So no GCP projects/billing accounts were ever deleted.
Is that accurate?
What is an expiration date if not a scheduled deletion date? Is it a soft-deletion? It's clearly ambiguous to me, who has no previous knowledge of this event before a minute ago. "Google auto deletes account at expiration date" seems to be a reasonable interpretation of what's been presented to me so far.
We should always maintain a healthy skepticism in either direction, and maybe supporting that is your purpose, but: why should your throwaway account be believed?
“During the initial deployment of a Google Cloud VMware Engine (GCVE) Private Cloud for the customer using an internal tool, there was an inadvertent misconfiguration of the GCVE service by Google operators due to leaving a parameter blank. This had the unintended and then unknown consequence of defaulting the customer’s GCVE Private Cloud to a fixed term, with automatic deletion at the end of that period.”
> Never forget what happened to Australian UniSuper
I don't know if media or the readers are at fault. The article doesn't even make sense.
> More than half a million UniSuper fund members went a week with no access
If Google really caused such a huge loss there would be no joint statement. The buyer i.e. UniSuper would be trying to sue them. The fact that it is a joint statement implies the two parties are sharing the responsibility. Now complaining about UniSuper is boring and so spinning it on Google Cloud gets clicks.
I remember once consulting one of my client who used GCP and to get SSH to one of the instances where pain in ass, I spent couple of hours on that.
If I recall correctly they do have concept of users and then user gets SSH key and you can't just put SSH key on instance out of the box as on any other VPS...
They make it pretty easy to SSH in through both the console and with the gcloud command. If you set it up correctly, Linux user accounts are tied to GCP IAM accounts and are automatically provisioned. It's actually super slick! Maybe your client didn't want to do that.
I'm surprised to find all comments so strongly against Google. I do not like their customer support nor their surprise product closure acts... but I do find their cloud interface easy to use, pricing is solid, and (K8s/Cloud Run) I have barely been affected by any significant outage in some 8 years of use.
Contrast to Azure who despite their SLA had significant outages every few months that were very noticeable, sometimes even requiring us setting up an entire new K8s project from scratch. AWS is better in terms of reliability. Their UX is not great though, I often feel like I'm a wizard waving my wand with permissions and connecting things like audit trails. Many actions are a bit delayed in their effect. I've also had to contact support at times to raise arbitrary limits in the platform.
In balance I would recommend GCP. Their product closures have not affected me enough to scare me away. I guess experiences vary depending on which product we are using.
i would say that smaller, independent enterprises/solo shops tend to not hate GCP.
Large shops - aka, a shop where you're not part of the controlling entity - tend to prefer either AWS, or Azure depending on how hard the CIO got wined and dined previously (and existing inertia - older shops that used AWS as an early adopter seems to stay with AWS).
> Their UX is not great though
i agree, but nobody really cares enough. On the other hand, i'd not pay more for better UX. Judging by the current state, i say a lot of customers fall into this category.
In quite a few places I saw, it would be either that or sheer incompetence (which was obvious to me at the time, and become obvious to people who came after me and finally decided to migrate away, with more losses on the way that could have been easily avoided).
You can't even set spending limits on Google Cloud and have to use some obscure script that deactivates billing if you want to imitate it. The whole product is incredibly buggy and complicated.
Wait, so you can now set up a hard cap on your spending and they guarantee they won't charge you more? It would be exciting but I don't believe they actually allow that especially as their competition also refused to implement that, in spite of it being no #1 requested feature for many years.
Okay, my bad, I was looking at this page [1] which says that the "Cost Management" feature is available for Pay-as-you-go subscriptions, but [2] is pretty clear on the fact that Spending Limits aren't available for Pay-as-you-go subscriptions.
So, TL;DR: Azure still doesn't allow you to set spending limits on Pay-as-you-go subscriptions
> AWS has slightly improved with their "budgets" and "budget actions", but it’s far from something I’d call a "spending limit".
Yeah they had to do something so they came up with a half-solution that basically informs you with a delay, doesn't prevent any quick damage and - if you decide to use budget actions - requires you to know what is actually causing the problem. Better than nothing I guess.
AWS is the best provider for tech people, the product is high level but versatile too. Price is somehow high, but the service is here.
azure is microsoft, so this explains that
gcp has "a vision": if you do not share it, it's a pain. Also, gcp has some sick design for core products : checkout load balancers if you want a "good" laugh. It's a stack of hack, put one on top of the others
I would not recommand against GCP, it's the average player: not the best, definitely not the worst.
This is an option, not a fact. Many of us don't agree. After working on all 3 I feel AWS is, by far, the worst experience for tech people. My perspective is that it's only doing great for execs and PowerPoint engineers.
My impression from a small project needing a OAuth scope was that the UI of GCP was sluggish, and although the other providers have sometimes sluggish UIs Google's was uniquely so. Anyone else feel this way?
Example is a sidebar that opens for adding an email during OAuth flow, after adding it clicking "Add" once did nothing, there was no feedback. Had to click it at least 5 times for it to go away and save correctly. In fact this is even specified in the documentation for the third-party tool (Google Drive downloader[1] step 18) that you have to click it multiple times. I don't think this is normal.
And also I should mention, depending on hardware specs the GCP portal was nearly unusable. So hopefully one will not need to rely on creating OAuth resources for a CLI program on the same computer. Although to be fair, I wasn't benchmarking against AWS/Azure on that hardware (since I needed to use GCP Google APIs).
Quite frankly, it’s hard to recommend any Cloud Service nowadays. Technical issues apart, the strategy _everywhere_ is to maximize picking the customers’ pockets. AWS is super expensive, Azure is super unreliable and GCP … well, it’s reliable yet features on it could vanish at a moments notice.
Most should just own their infrastructure, or use a more simple hosting service. Most companies could go very far with that.
Unless youf business is a global one with hundreds of millions of users in many jurisdictions, there is really no reason to go for cloud services. They tend to be much more expensive then the alternative, and have many ways to lock you in by making migration very expensive and time consuming.
For a medium size biz, I'd probably still recommend using GKE or EKS. A smallish EKS cluster of a few nodes will cost <1k a month for everything, it's fully managed so you don't need to pay anyone except for the initial setup.
Otw the cheaper operational cost would be to setup a few hetzner servers and manage your own infra via kubernetes or docker/ansible runbooks. But you should budget a few days each month for maintenance and general support, plus you'll need someone skilled in server maintenance to set it all up (becoming a rarish skill these days).
If on a ramen budget, you can set it up and let it rot until everything breaks, then migrate to EKS. :)
I generally view churn and nice primitives to be the essential balancing act for a platform. Perhaps any leading edge (software)system. We aren’t going to get anything complex perfect the first time, so it is change or mediocrity.
Funny. This is genuinely one of my deepest insights after 17 years in IT, some of it even as lead on infrastructure and platforms. Perhaps not well formulated, ok. And to date one of my most down voted comments on HN. What are the incentives supposed to achieve?
AWS has been feeling like the least terrible of the big 3 lately. Domain registration is the canary for me. Only AWS does it right. Azure has some approach but they force you into a lock in service and rely on GoDaddy for the actual registration.
These vendors underestimate the impact of having something so fundamental managed by a weird 3rd party. Route53 gives me a lot of confidence that I'm not going to be jerked around with domains and DNS. This factors heavily into my purchasing decisions.
178 comments
[ 4.4 ms ] story [ 215 ms ] threadYou shouldn't ask people to migrate before having a documented, tested, tool-assisted, clear and intuitive migration flow that takes very little time.
Maybe it is a strategy for cleaning up old apps or something, but I doubt it
At its core, alphabet is a services company, it does not sell goods (to the most part). Their board needs to decide if the company should focus on business services or consumer services and replace the executives with people who have experience in that area.
The actual quality of GCP is superb in my opinion. They have a flair for architecting excellent solutions. I prefer GCP over any cloud from a purely technical perspective. Their leaders just don't get that it isn't enough. They're not operating a toy or a museum of technical marvels. Actual people need to use it.
I saw tickets from high-paying, top-10 customers, go unanswered for days; no one senior enough on the team to answer felt it was more important than the day-to-day, and no one from the support/account executive staff felt they had the authority to demand it.
I see it otherwise. I think solving customer support is crucial to GCP's success, and since I agree with you that GCP is the better underlying product, there is a deluge of money passing Google by, just waiting for the right executive team to start caring about this to pick it up. Kind of like Microsoft pre-Satya.
* https://yosefk.com/blog/people-can-read-their-managers-mind....
After leaving Google I started working at a startup that runs its production environment on AWS. We debugged an issue and experienced exactly what you're describing: our MSK cluster refused connections. Eventually that was solved by rebooting something on their side. It took less than 24 hours from when we opened a ticket until the issue was resolved, with a few back-and-forth messages between us and the AWS support engineer.
I'n guessing that's because AWS support engineers live in eternal fear of ending up on a PIP if I rate them less than 5 stars.
They're a trash company, and anything you get from them, even when you pay for the storage, is at best accidentally delivered to you and they could roll over in their sleep at any moment and snuff it all out.
> "had to migrate my domain after Google decided to shut down Google Domains decided to shut down."
That reminds me in their interview process with me. So you are hinting they also use bots for hiring developers. Would explain everything.
Cut backs have finally forced AWS to sunset services, but they just hard refused for many, many years. People like using it because the workflows evolve, but AWS will not force migrations on you, sometimes to your own detriment.
When you first build an app, everything is fresh because developers haven't had to hamfistedly shove assumption breaking models and patterns into their program yet.
Cloud customers want to evolve to the latest tech, but redoing 20 years of the company for no business reason makes programmers happy but executives nervous.
Sadly, that seems to not be the case anymore, and it’s a worrying development. AWS giving up on their customer obsession makes it easier for the other CSPs, who aren’t as customer friendly, to compete.
From the outside I dont perceive the promotion dynamics for engineers working on old services is that different at AWS than Google or Azure. Finance and engineering do conspire to kill off things despite customer annoyance, perhaps as they should.
There is also a constant churn of retiring EC2 instance types, security certs, old database versions, EMR versions, OS versions, security practices, cost optimizations, etc to suck up your time with minimal business value. This is not different fundamentally than on prem, you just have slightly less control and an outside party forcing your hand to do the right thing isnt purely a bad thing...
AWS is pretty good about giving you a grace period to migrate off and informal warnings if they strategically want you to move to a different service of theirs before stuff gets killed if you are looped in with a TAM (e.g Data Pipelines vs more-expensive Glue). They seem to have recently migrated to a strategy where they disable services for new customers but actually dont kill them completely off but keep them on life support.
The only product I know well in your list is pre-VPC ec2 instances, which (to be fair) are a terrible product : it is cross tenant (as in : you can impact other customers). Good riddance.
I think many AWS services are well designed : isolated software-only components. They build a lot on top of a very stable infrastructure (VPC / s3 / ec2 / IAM), which means supporting a service is really cheap : they are just a couple of containers running somewhere
A company that maintains that many TLDs is expected to have long term commitment to the domain related businesses but it turns out to be just a fleeting interest.
I used Google Photos on the iPhone for a while though, it worked fine you just have to open their app so it syncs.
I get it, it was a "consumer" product essentially, hence selling the business to Squarespace instead of someone like Cloudflare. But anything related to DNS is going to make infra folk very wary of what else you might be willing to kill or neuter. It's just at that level of fundamental things that make operators skittish.
Grocery stores don't make their money from selling bread and milk, but a store that doesn't sell bread and milk is run by fools.
CF could probably get a lot more customers if they would allow you to use custom nameservers for your domain.
Why would they want that? The whole point of CF as a registrar was so you could use the other services. The registrar is sold at cost. It's a way to lock you in.
So that's not the use of the domains. It's to make it easier / lock in existing customers.
But they never marketed it, among all the backlash over turning down Google Domains there was never any true call-out to this. To this day I suspect many people don't know about Cloud Domains (which still exists and accepts new domains). I can't fathom any user-related reason for this - the contract in selling to Squarespace must have forced Google to be unable to properly market Cloud Domains as a transition option. If not, the Google Domains PM truly didn't know the existence of Cloud Domains?
> Caution: Transferring a registered domain from a third-party domain registrar to Cloud Domains is deprecated and removed.
So, maybe, it will be killed soon then.
Squarespace's domain panel crashed with a nondescript error when I tried to update nameservers prior to transferring out, and they shut off the Google nameservers as soon as the transfer went through on their side. To add insult to injury, Squarespace makes you wait 5 days for a transfer, with no way to expedite -- and in my case, they waited 6 days, taking me offline on a Friday night. This was the worst experience I've ever had using a domain service.
I also found GCP's developer tools and docs to be quite solid, better than AWS. Example: It's nice that they provide Terraform snippets for most of their resources.
Probably everyone. It’s the best solution for work emails, docs, calendars, etc.
It's the least worst solution for "whole-workplace" systems, but most people who are saying "won’t use Google for anything cloud" probably self-host or use something like PurelyMail (https://purelymail.com/) or FastMail (https://www.fastmail.com/) for their own emails
Why Hotmail? Because if the name isn't enough indication it's because I've had it for well over 25 years at this point and I like Microsoft anyway.
In the last 4 compagnies I worked for (the multi-billions kind of compagnies), everybody uses workspace, there are no office 365
- Does Chrome keep nagging you for login and bookmark sync, when you use O365?
- Google’s way of handling shared inboxes is awful (Want an extra email? Create a …Google Group! Then assign members, tune the perms, etc.) How does it work in O365? Can I just create an email and assign 7 users on it?
Anyway: big corps disable Chrome sync; most people will use Edge in enterprise because it integrates seamlessly into M365 and supports multiple work profiles (e.g. privileged accounts, service accounts, etc.)
Shared mailboxes are built into Active Directory; most enterprises have automated ways to create those -- usually the big concern is managing removing members when they're not supposed to be members anymore, hence the centralized management. There are also M365 groups you can create yourself, but that's just a distribution list. And generally, Teams is the way to go for any kind of collaboration work now - shared mailboxes seem to be a bit of a relic, usually meant for external interfaces that require an email address.
The URLs might be mentioned in many different places. Why break URLs just to make a "better product"? Why couldn't the new product keep old URLs working?
Unfortunately, they're also notoriously unreliable as a dumb pipe.
You're not wrong, but advertisers are a cancer on society that not only do not contribute any value but actively destroy the world around us. it's difficult to assign anything but deeply negative value to their needs and concerns.
"More than half a million UniSuper fund members went a week with no access to their superannuation accounts after a “one-of-a-kind” Google Cloud “misconfiguration” led to the financial services provider’s private cloud account being deleted, Google and UniSuper have revealed."
https://hacks.mozilla.org/2022/02/retrospective-and-technica...
“During the initial deployment of a Google Cloud VMware Engine (GCVE) Private Cloud for the customer using an internal tool, there was an inadvertent misconfiguration of the GCVE service by Google operators due to leaving a parameter blank. This had the unintended and then unknown consequence of defaulting the customer’s GCVE Private Cloud to a fixed term, with automatic deletion at the end of that period.”
https://cloud.google.com/blog/products/infrastructure/detail...
Are you saying that Google lied about being responsible for it? What would they possibly gain from that?
The original news article says Google deleted the account. As per your quote and official statement - no account has been deleted.
So that news article is completely false. Point proven.
Or did I get poe's lawed?
I don't know if media or the readers are at fault. The article doesn't even make sense.
> More than half a million UniSuper fund members went a week with no access
If Google really caused such a huge loss there would be no joint statement. The buyer i.e. UniSuper would be trying to sue them. The fact that it is a joint statement implies the two parties are sharing the responsibility. Now complaining about UniSuper is boring and so spinning it on Google Cloud gets clicks.
https://cloud.google.com/blog/products/infrastructure/detail...
Maybe the upside is that Google is in exchange able to offer:
> Google Cloud - great engineering, good product.
If I recall correctly they do have concept of users and then user gets SSH key and you can't just put SSH key on instance out of the box as on any other VPS...
Contrast to Azure who despite their SLA had significant outages every few months that were very noticeable, sometimes even requiring us setting up an entire new K8s project from scratch. AWS is better in terms of reliability. Their UX is not great though, I often feel like I'm a wizard waving my wand with permissions and connecting things like audit trails. Many actions are a bit delayed in their effect. I've also had to contact support at times to raise arbitrary limits in the platform.
In balance I would recommend GCP. Their product closures have not affected me enough to scare me away. I guess experiences vary depending on which product we are using.
Large shops - aka, a shop where you're not part of the controlling entity - tend to prefer either AWS, or Azure depending on how hard the CIO got wined and dined previously (and existing inertia - older shops that used AWS as an early adopter seems to stay with AWS).
> Their UX is not great though
i agree, but nobody really cares enough. On the other hand, i'd not pay more for better UX. Judging by the current state, i say a lot of customers fall into this category.
A decision maker can basically choose from where they want to get bribed, so it won't be a crucial factor most of the time.
Full downtime for 2 days, no apologizes, no answers, had to pay 3% extra for support, no credits.
Special award to the dude representing Google trolling on Twitter that they will do another "wild Friday release".
Shitty.
I wanted to link the specific article mentioning that, but it seems like this works now :)
I haven't tried it myself, and I hope it's true (:
So, TL;DR: Azure still doesn't allow you to set spending limits on Pay-as-you-go subscriptions
[1]: https://learn.microsoft.com/en-us/azure/cost-management-bill...
[2]: https://azure.microsoft.com/en-gb/support/legal/offer-detail...
AWS has slightly improved with their "budgets" and "budget actions", but it’s far from something I’d call a "spending limit".
Yeah they had to do something so they came up with a half-solution that basically informs you with a delay, doesn't prevent any quick damage and - if you decide to use budget actions - requires you to know what is actually causing the problem. Better than nothing I guess.
azure is microsoft, so this explains that
gcp has "a vision": if you do not share it, it's a pain. Also, gcp has some sick design for core products : checkout load balancers if you want a "good" laugh. It's a stack of hack, put one on top of the others
I would not recommand against GCP, it's the average player: not the best, definitely not the worst.
This is an option, not a fact. Many of us don't agree. After working on all 3 I feel AWS is, by far, the worst experience for tech people. My perspective is that it's only doing great for execs and PowerPoint engineers.
In opposition with people who enjoy taking a product of the shelf and call it a day
But you are right: this is my opinion, not a fact
Example is a sidebar that opens for adding an email during OAuth flow, after adding it clicking "Add" once did nothing, there was no feedback. Had to click it at least 5 times for it to go away and save correctly. In fact this is even specified in the documentation for the third-party tool (Google Drive downloader[1] step 18) that you have to click it multiple times. I don't think this is normal.
And also I should mention, depending on hardware specs the GCP portal was nearly unusable. So hopefully one will not need to rely on creating OAuth resources for a CLI program on the same computer. Although to be fair, I wasn't benchmarking against AWS/Azure on that hardware (since I needed to use GCP Google APIs).
[1] https://github.com/glotlabs/gdrive/blob/main/docs/create_goo...
Customer obsession (ugh) is just not in their DNA. I can tell firsthand they care more about what they want to do than what the customers want.
Where might one go?
Most should just own their infrastructure, or use a more simple hosting service. Most companies could go very far with that.
Unless youf business is a global one with hundreds of millions of users in many jurisdictions, there is really no reason to go for cloud services. They tend to be much more expensive then the alternative, and have many ways to lock you in by making migration very expensive and time consuming.
Otw the cheaper operational cost would be to setup a few hetzner servers and manage your own infra via kubernetes or docker/ansible runbooks. But you should budget a few days each month for maintenance and general support, plus you'll need someone skilled in server maintenance to set it all up (becoming a rarish skill these days).
If on a ramen budget, you can set it up and let it rot until everything breaks, then migrate to EKS. :)
Unless there is a team of 20+ engineers, I cannot recommend Kubernetes to anyone. It is a great primitive that's too primitive for smaller teams.
When your needs are low, AWS and friends are very competitive: quicker deploy, no CAPEX, lower OPEX
If your needs are high, you should indeed build your stuff on top of instances
If your needs are higher, then rent rack space and build on top of baremetal
If your needs are even higher, then build your datacenter
As an example, I put my personal backups in the "cloud". Is 1€/mo for 100GB cheaper than using a VPS ? Hell yeah !
These vendors underestimate the impact of having something so fundamental managed by a weird 3rd party. Route53 gives me a lot of confidence that I'm not going to be jerked around with domains and DNS. This factors heavily into my purchasing decisions.
Azure I can't use at all. Impossible to find the products I want in the UI.
https://news.ycombinator.com/item?id=32547912
https://news.ycombinator.com/item?id=33737577
https://news.ycombinator.com/item?id=35133917
https://news.ycombinator.com/item?id=22146082
https://news.ycombinator.com/item?id=22705122
https://news.ycombinator.com/item?id=4013799