4 ms·

I highly recommend Kubernetes as infrastructure. It has a reputation for being too complex to use on your own or with simple projects but that reputation is und

I highly recommend Kubernetes as infrastructure. It has a reputation for being too complex to use on your own or with simple projects but that reputation is undeserved. Self-healing container orchestration has been eye-opening for me. Many people groan at the prospect of learning something new but it is remarkably easy to use, the only barrier to entry being the high cost of cloud solutions and the unwillingness of many engineers to work with hardware (which would nullify the cost of cloud services). You can easily develop and test on local hardware and deploy to the cloud with the exact same configuration. The idea that your server does not perform regular health checks or spin itself back up when it fails just seems weird to me now. I like being spoiled.

simmanian 7y ago
I am leaning toward learning k8s seriously and am actually curious on your take: is the overhead of learning and maintaining a k8s cluster actually better than using AWS features like autoscaling coupled with health checks?
- ShakataGaNai 7y ago
  Don't maintain the cluster. Have someone else run it for you. Unless you want K8s cluster to be the only thing you do. The advantage of K8s is that it abstracts so much away from you, that you should (in theory) be able to take the same YAML config file from AWS EKS to GCP GKE to Azure AKS...and it runs the same everywhere. Things like loadbalancering and HTTP ingress rules that normally would manual configs on each platform - become part of the K8s config.
  - xur17 7y ago
    > Don't maintain the cluster. Have someone else run it for you. Unless you want K8s cluster to be the only thing you do. This can be as simple as just using Google's GKE service.
- gatherhunterer 7y ago
  I find my home cluster to be laughably easy to set up, and I am still new to home clusters. This is an area of rapid development. I could wipe the hard drives and have it back up in less than 30 minutes, not counting the time to rewrite the OS images. I just have some machines running Ubuntu Server with one running Ubuntu Desktop (just because I like it). Each node needs a single shell command and it’s good to go. In the past I have done things like using interface forwarding to allow every node to use a hard drive that is connected to the master so if I ever actually have a use for that or if I want the master to act as a DHCP server and forward its network interface to the workers I can do it again. To maintain portability you need to pay attention to CPU architecture and the underlying OS. Just make sure that your images are built for the same architecture that you will use in the cloud and that you have the right OS and version. x86_64 is ubiquitous (I think AWS offers an ARM server option, though) so developing on a Raspberry Pi cluster might be a headache to deploy. For production you will want to use a cloud provider (or not, it’s not the craziest idea to run a home server if your use case permits) and you can do so for development if you can afford it. The real advantages are service discovery, monitoring options, stateful support and resource allocation. I used Kubernetes Up and Running 2nd Edition, and if you don’t know Docker yet I would work through the Docker Book first because you will want to write your own images. Just make sure that everything lives in the same repository: a single filesystem as the single source of truth.
ShakataGaNai 7y ago
Having run K8s a bit on my own... Please don't run your own cluster. K8s is great and has a lot of wonderful features which help immensely with automation (ex: You tell it "I want to run 3 copies of this software and here's how to check it's health" and if it sees unhealthy containers or less than 3 copies running... it fixes the problem). However K8s is a bear to run in its own right. As a single-person company you don't have time for that. Use GKE or similar.
- gatherhunterer 7y ago
  You are misinformed. MicroK8s and K3s can get a full-fledged Kubernetes setup running in less than a minute. I run a five-node cluster at home and it is a breeze. Edit: I forgot about HyperiotOS, which is another viable option. MicroK8s has Canonical’s backing so it has quickly become the most well-supported option. It advocates single-node clusters for development but it supports multi-node as well.

simmanian 7y ago

I am leaning toward learning k8s seriously and am actually curious on your take: is the overhead of learning and maintaining a k8s cluster actually better than using AWS features like autoscaling coupled with health checks?

ShakataGaNai 7y ago
Don't maintain the cluster. Have someone else run it for you. Unless you want K8s cluster to be the only thing you do. The advantage of K8s is that it abstracts so much away from you, that you should (in theory) be able to take the same YAML config file from AWS EKS to GCP GKE to Azure AKS...and it runs the same everywhere. Things like loadbalancering and HTTP ingress rules that normally would manual configs on each platform - become part of the K8s config.
- xur17 7y ago
  > Don't maintain the cluster. Have someone else run it for you. Unless you want K8s cluster to be the only thing you do. This can be as simple as just using Google's GKE service.
gatherhunterer 7y ago
I find my home cluster to be laughably easy to set up, and I am still new to home clusters. This is an area of rapid development. I could wipe the hard drives and have it back up in less than 30 minutes, not counting the time to rewrite the OS images. I just have some machines running Ubuntu Server with one running Ubuntu Desktop (just because I like it). Each node needs a single shell command and it’s good to go. In the past I have done things like using interface forwarding to allow every node to use a hard drive that is connected to the master so if I ever actually have a use for that or if I want the master to act as a DHCP server and forward its network interface to the workers I can do it again. To maintain portability you need to pay attention to CPU architecture and the underlying OS. Just make sure that your images are built for the same architecture that you will use in the cloud and that you have the right OS and version. x86_64 is ubiquitous (I think AWS offers an ARM server option, though) so developing on a Raspberry Pi cluster might be a headache to deploy. For production you will want to use a cloud provider (or not, it’s not the craziest idea to run a home server if your use case permits) and you can do so for development if you can afford it. The real advantages are service discovery, monitoring options, stateful support and resource allocation. I used Kubernetes Up and Running 2nd Edition, and if you don’t know Docker yet I would work through the Docker Book first because you will want to write your own images. Just make sure that everything lives in the same repository: a single filesystem as the single source of truth.

ShakataGaNai 7y ago

Having run K8s a bit on my own... Please don't run your own cluster. K8s is great and has a lot of wonderful features which help immensely with automation (ex: You tell it "I want to run 3 copies of this software and here's how to check it's health" and if it sees unhealthy containers or less than 3 copies running... it fixes the problem). However K8s is a bear to run in its own right. As a single-person company you don't have time for that. Use GKE or similar.

gatherhunterer 7y ago
You are misinformed. MicroK8s and K3s can get a full-fledged Kubernetes setup running in less than a minute. I run a five-node cluster at home and it is a breeze. Edit: I forgot about HyperiotOS, which is another viable option. MicroK8s has Canonical’s backing so it has quickly become the most well-supported option. It advocates single-node clusters for development but it supports multi-node as well.