8
Comments
  • 0
    Nah let’s throw loads of money to anything people are talking about right now and see what happens.
  • 0
    Digital Ocean is a "cloud" and it is just a machine. Give me any system running on windows, Mac or Linux and I can run my applications without any change in code.
  • 3
    Ideally - yes. But having provider-speciffic implementations at the lowest layers of your app arch are performance gainers.
  • 0
    Not unless you want to add a bunch of complexity for very little return.

    Pick a cloud and use that, every cloud has proprietary things that might not translate 100% to another cloud.

    If the cloud you pick is AWS then you have the benefit of whatever problem you are trying to solve - it's almost guaranteed someone with more money than you already had that problem and its been solved in the platform already
  • 0
    @unsignedint you're screwed if AWS change their pricing or deprecate something you're dependent on though.
  • 0
    You can go a long way with some solutions (kubernetes for example) but unless you want to maintain everything yourself you have m rely on services. As mentioned on some layers there will most likely always be some vendor specific stuff going on (storage, loadbalancer/assignable IP.

    I am a bit disappointed with the lock-in of cloud functions though. There is a solution for deployment but the code itself must be modified a lot.
  • 1
    @hjk101 For Kubernetes there are mixed results for PaaS offerings dependent on your provider. AKS on Azure seems pretty decent, EKS on AWS less so as Amazon openly admit. I've not tried Google's offering but their data agreements rule Google out for us anyway. But even using AKS you're tied to Azure and dependent on them. If Azure become unfavorable for whatever reason or lose compute, it's not an easy job to change provider quickly or automatically.
  • 0
    Vendor lock-in can get very expensive in the long run. Technical debt, if you will.
Add Comment