DO Ideas 2

Implement auto-scaling of resources

Use case:

I have a $10/month VPS and one of my blog posts gets posted to Hacker News and Slashdot. My server sees a huge spike in traffic. This is okay because Digital Ocean has automatic resource scaling and they will bump me to a higher plan (with more ram, cpu, etc) for as long as my server needs the extra resources to handle the requests. When my blog post dies down, I return to my $10/month plan and I get charged based on the amount of hours I needed to use the larger plan.

  • Max Mackie
  • Sep 11 2018
  • Sep 11, 2018

    Admin Response

    We are gathering information about autoscaling needs, especially given the introduction of our Load Balancer product. Your feedback is welcome, especially if you can describe your use case and tools you are using to automate workload deployments.
  • Attach files
  • Ben Yanke commented
    September 11, 2018 19:24

    I agree, this is important - and especially given there is already a set of monitoring tools and alarms, probably relatively trivial to build.

    Just tie alarms to scaling events, add a few ancillary config knobs, and give a provisioning script or image to start new images from.

    boom - done.

    Heck, I could even do that outsourced, if monitoring data was available in the API or by webhooks....

  • Anonymous commented
    September 11, 2018 19:24

    Would you be able to come up with a feature where we can upsize our droplet automatically as our traffic grows on our website. This will be a handy feature on the cPanel.

  • Olaf Lederer commented
    September 11, 2018 19:24

    I think this feature should work like the one vps.net offers
    http://wiki.vps.net/vps-net-features/power-pack-faqs/

  • Moisey Uretsky commented
    September 11, 2018 19:24

    We do not have any plans currently to implement auto-scaling mainly because there are two issues.

    One is that auto-scaling a particular droplet depends on the service that is running, whereby many customers that use it to run their DB + web server may not want to incur a 2-3 minute downtime while it is being resized.

    The second is that alot of this can be achieved through a distributed layout of your application with open source tools like puppet and chef where new web nodes are launched automatically through our API.

    Thanks

  • Kroeze Jordy commented
    September 11, 2018 19:24

    Would be nice, but ofcourse with the option to do this or not, and limit what will be the max and min plan.