$2000 I could see, but I do think more people are overspending on CPU for gaming PCs by probably double. Maybe not overspending if you're getting a tangible productive value out of it, but I suppose they didn't specify gaming. For gaming specifically I'd probably try and balance the useful GPU power I need with the minimum necessary CPU to prevent bottlenecking.
Part of the reason I haven't upgraded my intel macbook pro, is just because I think the cost all-in seems outrageous, even for someone who exclusively works on mac. I can't rationalize $500/16gb of ram or w/e. I haven't upgraded my gaming PC much, because the cost of GPUs very quickly overwhelms the performance improvement I'd get value out of compared to my severely out of date gear that I found by the roadside.
Just an anecdotal counter point: very happy with gcp.
Best network from all and coherent modern ui.
Not the usability hell like azure... (You know when clicking on a often used resource on the start page which let's you jump directly to it but doesn't allow you to jump a level up of all the other resources of the same type which totally works fine when you navigate to it the normal way... Or the huge hassle and complexity of resource groups for f everything...)
But you know the tweet not even states what quota was reduced.
Azure's support has also been fantastic. Depending on priority of ticket, with an enterprise account things get picked up sometimes minutes after raising them.
Their AKS offering was a crap show during the first year of general release and I opened countless tickets and they were snappy at the response times which is interesting seeing at the scale they operate.
Judging by all the comments GCP seems like one to avoid? Which is a shame because I had a desire to train multi cloud. If they treat their support like the rest of their products then I'll advocate a different IaaS where I'm able to.
I think most of the comments in this thread are probably about the lower support levels for GCP. I personally have never used enterprise support for GCP. For the lower support levels AWS is miles better than GCP. Not sure about the enterprise support levels. For the hobbyist/early stage, AWS support is the top pick for sure.
I read in a much older thread that GCP’s TAM and enterprise support is pretty good:
> Hey, thanks for all that you’ve done. My experience with GCP has been an incredibly positive one. GCP documentation has always seemed fantastic. Our TAMs were very responsive.
> GCP support has by far been the best support experience. I have to say that the initial days it seemed to suck. The UI was some 90s google group clone which wasn’t even accessible through the GCP console, it was its own separate site which I always found amusing. But over time, the UI and quality of support became more streamlined and predictable, and I consider it one of the best SaaS support experiences today.
> One particular incident I’ll never forget is a support person arguing with me why network tags based firewalls are better overall for security than service accounts based firewalls. I expected to have a very cut and dry exchange but the support engineer actually did convince me that tags are superior to using service accounts. I did not ever expect to have had such a discussion over enterprise support tickets.
I would smash my head in a door before being reliant on google and their (lack of) support for any business critical services. It's fine as a secondary, but my god would it make me really nervous if that was the primary cloud platform.
For what it’s worth there is a tabular view that has more item detail. Things like marketplace subscriptions can often appear as Other in the graph.
I’m pretty happy with azure. I’ve not used GCP but vs AWS I find permission management a lot nicer. I have yet to have a billing surprise that didn’t come down to me not reading closely enough.
What I don’t like is they sometimes lock private links (aka the ability to not have a service publically routable, only accessible on a vnet) behind premium SKUs, looking at you service bus.
Do you have any links to more info about this?
I don't use Azure, so I'm not sure where I would look for something like this.
I've been aware of some enterprise projects that are running on Azure, where from the description of how their architecture worked, I couldn't understand how they weren't drowning in network cost. It makes a bit more sense now if I know that Azure weren't charging people for some types of network bandwidth.
This is certainly a troll post. I use all three cloud providers. GCP is the worst of all, just try their simple text to speech UI. It doesn’t work most of the times.
Don’t even get me started on a deployment story for GCP their deployment manager is deprecated and redirect you to use terraform.
I hate to swallow it but Azure was more usable and straightforward.
Azure has had multiple global outages for many services. I believe you can look at overall stability and see it's not in the same league as GCP and AWS.
Lol can you use ed25519 keys with Azure yet? I doubt it. It was and will continue to be a joke because MS can't pay or promote people that care before they leave.
Azure’s target customer group is non-tech enterprise. The type of group who generally buys solutions over builds them in house. Where needing ed25519 keys is not a common ask
Gcp used to have a big network offering lead. But not anymore. AWS has a more reliable network (hello no global outages) and there’s not the same performance gap since AWS and azure’s investments in private fiber connectivity panned out.
A UI for cloud services is super helpful to have when exploring, troubleshooting, and noodling around. But yeah, massive red flag if anyone’s using it to deploy production services.
That’s a pretty myopic view of what a UI can be used for. I’m all in on IAC but still like having a UI to click around to observe state rather than memorize a bunch of CLI commands.
And even if I can get what I need via CLI it's nice that higher level people in the org can click around and get what they need without having to bother me to retrieve the data for them.
I’ve found no megacorp makes a good UI for their products that speaks to my infra. Frankly I usually use something like Grafana and write my own TUI for the rest ;-) but there are tons and tons of infrastructure management and visualization software products out there. Generally I look at cloud providers for their api, capabilities, etc. Their UI ability is so far down the list that Id rather buy someone else’s for purpose single product strategy than the “oh yeah the console” version the providers must be providing given the economics of things.
I think they may be being downvoted not because their opinion is different but because usability of the web UI is possibly the least important attribute in choosing a cloud provider for a lot of people given the use of tools such as terraform.
They replied to an incident report with "works for me". (With an implied "sounds suspicious" in the last sentence, but I don't pretend to know whether that's what they meant.)
OP is being downvoted because his anecdotal defense of GCP due to nice UI is irrelevant to the link posted. And probably to cloud computing in general.
Well, part of his criticism is that the UI is woefully inconsistent with regards to "upwards" navigation. If you navigate to a resource via the resource group, you can navigate back "up" to the parent RG. If you nagivate to the resource from the home screen, IIRC, you cannot. That's the criticism.
Yes, RG's are a pretty killer feature, and trying to understand the organization of resources in GCP is hard by comparison, and an utter nightmare in AWS. I'm not sure why he's knocking that. (And … GCP requires projects, which seem equivalent to the complaint against RGs.)
… that said … there are so many other things utterly and horrifically wrong with Azure that I wouldn't put them on a pedestal for resource groups. As much as I do like RGs.
(Also, "A resource group is literally just a folder" sigh, no, because they're not hierarchical. Azure goofed hard there.)
Resource Groups are supported pretty well within the CloudWatch ecosystem (you can create an AppInsights application from an RG and you can filter CW alarms by RG)
Blocking UI access in prod is fine, but developers should be able to experiment rapidly and in an agile manner in dev environments. If you forcibly block UI access in those cases, developers simply do less exploration and stick to tried-and-true solutions, even when those solutions no longer suit the problem at hand.
Azure has this too, you can have what they somewhat confusingly call subscriptions, which are logical units with their own billing, limits and permissions inside and organizations account.
I am not 100% sure how IPFS works, but it's where some NFCs are stored, right? Is it even possible to only store blobs you care about? Either way, it does solve the migration problem.
And reporting to the vendor is suicidal. At least assuming the stories I hear about vulnerability disclosures are representative, which I think they are.
In their place, if I were to inform the company, I'd do it anonymously. If it was an actually important issue - as this very much looks like - I'd consider informing the building manager, HOA, the gas installation company they use, and every local journalist, all together so they know about each other - and then CC that to the vendor.
Another option can be your country's CERT. In reasonably developed countries they generally have competent enough people to understand the concept of responsible disclosure (i.e. won't try to harass you for doing a good thing), and if they realize "oh shit, this is a critical infrastructure risk" they're probably in the best position to address not just the specific case, but also drive improvements (including via regulation) across vendors.
Crypto was advertised as decentralized and unregulated.
Most arguments I had with cryptobros were some sudo arguments about this stuff instead of just omitting that it's all about speculations.