Hacker News new | past | comments | ask | show | jobs | submit login

I think this article is comparing apple and pears to some extend. My company is solely cloud based and are using Azure, AWS and GCP, although 98% of our bespoke infrastructure runs on AWS. We could migrate this infrastructure to GCP, as Google has been catching up with AWS on cloud infrastructure, however, I would dread to migrate it all to Azure. We mostly use Azure for Azure AD (and Office 365), and AWS to build out APIs.

Basically we are using Microsoft for out of the box services and AWS for anything else.




As an engineer working on Azure, why would you be opposed to migrating everything to Azure specifically?


Not OP, but want to chime in.

My experience with Azure support has been dreadful, to say the least. Insanely long response times, being bounced from agent to agent, people just misreading the ticket. I really don't want to deal with Azure at this point because I don't trust the support.

Most of our company infra is on AWS, and while they aren't without fault, I can say that every time we needed to talk to a human being from their side, it was quick and on-point.


To chime in with another single data point, my experience with Azure support has actually been pretty good. Tickets take anywhere between 2h and 72h to get a response, which I'll grant is not ideal, but the ticket handlers I've dealt with have all been knowledgeable and helpful.

I work for enterprise-level companies though, so I don't know if support varies for smaller ones depending on spend.


I agree. I work for a company who is a top 100 globally customer of Azure. In case where we have needed support, depending on how severe we define the issue, we are contacted within minutes. Well within the defined contact times stated as when creating support tickets. In several other cases, Microsoft has brought in people from around the world to help workout various issues. However, I don't know if the response times and help offered would be the the same if we were a different scale organization.


I'm just a private individual hosting his blog on Azure, and I've had a stellar experience. I made a post about some gripes I had setting up my website, and a person from the developer outreach program reached out to me and put me in touch with the engineering team directly. Turns out my main complaint (lack of support for managed SSL cert for root domains) was already in the pipeline. They even sent a bunch of MS/Azure gear to my doorstep as a thanks for my feedback.


Do you think you know or could guess an average response time for ticket responses?


We don't open enough tickets for this to be really meaningful (things generally work as expected), but I'd guess around 8h


That's on point, I used the Azure platform for around two weeks to see if it would be worth for our company to move to it and my account got randomly banned, no answer from support except that they don't want to say why. I'd not trust Azure to put anything important on it based on my past experience.


I work in a SaaS company, and we run "a lot of Azure". We open around a dozen tickets annually. The response is a mixed bag. The response times are pretty good, but it depends what will you get as a response. Sometimes you get asked a bunch of irrelevant questions just to get a simple answer (which actually helps). Sometimes you just can't get them on the same page. There are support engineers that just do not know what are you talking about. Typical example: snapshots. It took me once like 10 emails to clarify that I'm having issues with managed disk snapshots, not with blob snapshots. At the end I'm not even sure he was still aware what am I asking, the guy was completely oblivious. He finally sent me some screenshot of his own control panel (where he sees much more info than the user does), to show me something, where I actually noticed something else that unstuck me. But if he didn't we would still be sending mails back and forward.


Not OP but, your global outages freak me out. I moved companies close to a year ago and with that platforms that I develop on (from Azure to AWS). Your global outages of which I can't engineer around at all don't give me faith in your platform.

Whereas AWS, sure, service outages occur but they're often region specific which I can build around.


Not OP, but in a similar boat. Basically, having everything in one cloud provider's hardware is a risk to uptime guarantees is the biggest reason. There are other, less good reasons too, but that is the one that can't be changed by adding features to Azure (or any other).


Is Azure unique in that if you shift to Azure you cannot also use other cloud providers but if you are on GCP or Azure you can use the other?


AWS has better open source managed solutions than Azure and GCP right now, but Azure also has things like a billing console that can integrate with AWS - which is a great feature albeit not being so powerful yet.


Azure is still catching up in many areas. For example low-priority VMs is a recent addition to Azure, whereas AWS has had these instances since 2009. I also find the Azure interface really clunky compared to AWS and I am definitely not a fan of the AWS console either.


Azure and GCP had either comparable or worse UIs for a lot of services compared to AWS, but region specific UIs were a notable unintuitive design compared to others.


How different are the APIs between Azure/AWS/GCP? How easy or difficult would it be to make that switch?


Very different, at least between Azure and AWS. The Azure one is very "manual", you must do a bunch of operations separately which are abstracted as a one-click-thing in the portal. Also, when working through the API, you get a sense of how unreliable some processes are. Of course you should have circuit breakers, failovers, rollbacks and stuff. But with Azure API, they are a MUST.

We also work with the GCP API and it's somewhat closer to Azure, AFAIK (haven't worked with it directly).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: