Go look up how to create a load balancer on gcp. Last I looked there were six pages none of them simple and clear. I just want to know how to create one. It’s one simple thing. The essential part of any cloud service.
Now go read the aws documentation on an ALB. One page, this is how you create one and this is how you use it.
Azure on the other hand arrived late with a spectacular business plan. “You already pay me for software, cloud is part of that”. They won enterprise cloud in the blink of an eye.
Meanwhile Gcp wrestles with writing task oriented documentation, easy things are obscure or weird. (Compare security groups at aws with gcp, now look again. You didn’t get it the first time ) also if you’re designing a service there’ll be some critical thing that just doesn’t work. The old thing is deprecated the new thing is beta. (You fool! Don’t use beta things in production! Oh sorry there’s the non beta thing I can use? There isn’t one) or the feature you need to complete your task will never arrive because there’s a pissing march between gcp and k8s over who’s job it is to make the ingress work as desired.
Also, as others have said their support is garbage (I found a customer evangelist who was amazing if I could wait 3 days to get a response)
Google doesn't understand Enterprise. They expect enterprises to use features or solutions that they think is the best way or highway. AWS/Azure bends their back to introduce whatever features enterprise needs.
Next is poor support. I worked in a team that used DialogFlow enterprise for a voice app. The NLP engine was way way better than Alexa. But if you face an issue their enterprise support will tell you to post in StackOverflow !! As an example of why they don't get enterprise, DialogFlow didn't support mTLS for authentication and had no plans to build that feature.
I think also: AWS started the industry so first mover advantage and a hell of a lot of products in the space and experience. Azure (MS) has every corporate in the world (nearly) using is on prem AD so they would have picked up tons of easy Azure clients just extending their corporate networks to Azure AD.
This is much bigger than any "google killing products" issue. Azure got a lot of ways to push itself in, and reaps a lot of possible clients by being first party windows solution for companies that have a lot of windows stack.
AWS has the first mover advantage and enormous brand penetration - it's literally the "you can't get fired for choosing AWS" (instead your company might fold due to AWS bills, but you'll be clean).
Meanwhile Google seems to be much less known in cloud infrastructure space - people think of Google Apps/Workspaces, not GCP, and they seemed to have serious problem getting to clients. I know I was quite surprised by the available stuff and low prices when I did my first project on GCP back in 2016, and generally I've been pleasantly surprised except by their horrible billing support (Google can't into actually getting paid?)
They are in this funny situation where they constantly seem to be lacking fully blown virtuous cycle of adopters and supporting companies, despite having some of the best options on the market - but I think the real issue is that they are neither the first mover (AWS) nor "canonical windows source" as with Azure.
I agree with the first point but can you cite some examples for the second. I personally feel they are/should be most developer friendly company considering they are purely engineering driven.
A lot of it seems to be the vastly different quality of documentation and support based on programming language. Where AWS seems low level enough it doesn't matter what language you write your software in, and Azure has made "bring your software no matter what language you write it in" a very explicit message even in its high level cloud offerings, Google gives the strong impression/ego that there are "right languages" and "wrong languages" and they aren't going to support you well (or possibly at all) in the "wrong languages".
Some of that impression is leftover from the way App Engine handled things and the first impression that left, but even GCP documentation itself still seems to struggle outside of "Google approved" languages.
Disclaimer: I interviewed with GCP as a .NET developer with a lot of experience in .NET to try to improve the situation. I got a lot of nasty ego directed at me that interviewers didn't trust my technical expertise because it was in such a "gross" stack and didn't know how to interview me technically. I personally saw that as direct evidence for why I'd never use GCP.
Why would the one imply the other? The Spartans were the most military driven nation in history (possibly excepting the Mongols, who are another fine example) but they weren't exactly open-arms to foreign soldiers..
1) GCP support is non-existant
2) Google has a track record of dropping services with little to no notice.
3) There are numerous stories of google locking or deleting gcp account because the a.i. misinterpreted something about the owners gmail account.
4) Try to get a human on the phone, its impossible.
It doesn't matter how good the services or pricing are. I dont trust them and I doubt I ever will.
I like GCP quite a bit, and I've architected on it both in the "giant global company" and "small startup" spaces, but it's just not well-positioned competitively. AWS has first-mover advantage and massive lock-in, while Azure has a comfortable second place position with companies that are deeply ensconced in the MSFT and .NET ecosystems and took longer to build a cloud strategy. Meanwhile, GCP was well below feature parity for a long time, didn't really have an enterprise strategy that would allow it to take on its competitors, and never successfully positioned themselves as the startup cloud of choice despite that being their focus.
Today, GCP's still better-positioned for small companies who need to move fast and are more price-sensitive, but they've done a good job catching up on the enterprise space, and having late-mover advantage has helped them avoid some of the footguns you see in AWS. There are still a lot of sharp corner cases in their services and documentation, and they're as bad as anyone else in Google when it comes to taking customer direction (it really helps to have direct contacts in the organization or be a large-scale implementation partner), but they're a perfectly valid cloud option with a lot of great services at relatively aggressive price points (and, arguably, if you're working in health or life sciences, they're actually a very good option compared to AWS or Azure thanks to their extensive healthcare API portfolio). Unfortunately for them, they're never going to be anything more than the fourth-largest global cloud provider, which isn't a bad place to be, but probably a bit humiliating for GOOG.
IMO, lot of services are not matured enough. For eg: Changing roles or even a security group from GKE/Nodepool requires nodepool replacement. We need to take care of the complete node draining and adding the new nodepool.
I too agree that GCP doesn't take customer feedbacks like AWS do.
We can add new rules to existing network tags(security groups). However, Adding new network tags to node pool require node-pool replacement at this time. That's pretty wierd.
I had several thousand dollars in GCP credits but eventually moved back to AWS.
Overall I prefer the philosophy of AWS regions vs global infra. So many GCP outages were global. Also AWS seems very keen on customer feedback (if you are big enough at least).
Slower to market than AWS. Doesn't have their foot in the door as well as MS (due to Win, Office, etc). Never made any real push, or unique value proposition, to meaningfully break in.
I won’t touch GCP, even after they gave us 2k credits. Their customer service reps have “well that’s just too bad” on the tip of their tongue. AWS is #1 for good reason
Now go read the aws documentation on an ALB. One page, this is how you create one and this is how you use it.
Azure on the other hand arrived late with a spectacular business plan. “You already pay me for software, cloud is part of that”. They won enterprise cloud in the blink of an eye.
Meanwhile Gcp wrestles with writing task oriented documentation, easy things are obscure or weird. (Compare security groups at aws with gcp, now look again. You didn’t get it the first time ) also if you’re designing a service there’ll be some critical thing that just doesn’t work. The old thing is deprecated the new thing is beta. (You fool! Don’t use beta things in production! Oh sorry there’s the non beta thing I can use? There isn’t one) or the feature you need to complete your task will never arrive because there’s a pissing march between gcp and k8s over who’s job it is to make the ingress work as desired.
Also, as others have said their support is garbage (I found a customer evangelist who was amazing if I could wait 3 days to get a response)