The way I look at the cloud (why I love Rancher, and what I hope to see next)

The cloud is an incredible way of segmenting resources into pools of disprate type. Just for example, when I see a cloud, really I know that I have aggregate pools of:

  1. Compute
    Raw CPU & GPU Cycles.
    I love that Rancher lets me use these much more efficiently.
    I imagine that in the future, Rancher will be able to migrate containers for me between hosts to keep performance up and cost down.

  2. Memory (RAM)
    The aggregate pool of RAM resources on my clouds.
    My RAM utilization goes up with Rancher, instead of sitting at a rather low % of total.

  3. Network Availability
    The IPV4 and IPV6 addresses and pool of ports in total that I have available to me.
    I’m hoping to manage these addresses programmatically with Rancher in the future.

  4. Network Capacity
    The inter-node bandwidth available to me, the internet bandwidth available to me, and the per-customer-channel (“delivery”) bandwidth available to me.
    I’m hoping to get some reporting on my network capacity performance information from Rancher in the future

  5. Storage
    Today, as far as I know, Rancher doesn’t do too much with Storage. I would love to see it pool the storage of the various hosts that I connect to it, as well as support Google Cloud Storage/S3/NAS across all of the nodes.

What Rancher’s Doing for me

It bridges intentionally and unnecessarily separated resources that are separated to serve the interests of others (that is to say, public clouds). By pooling them, it enables me to manage the resources on my own. By letting me easily use and visualize containers, Rancher more or less taught me cloud management and gave me new insight into how web applications can be composed.

My most-wanted features

  • Storage Aggregation between my existing nodes
  • A way to use a service like google cloud storage / Amazon S3 as an addendum to my server’s hard disks
  • Support for LXC, because stateless containers are just not enough
  • RancherOS + Rancher on armhf + arm64 (working on that over here, but just like you, other things keep taking priority, and those other things certainly aren’t bad)

Thanks guys!

-Jake

1 Like

On wanted features - we would love to see detailed metric reporting.

We plan to run the infrastructure for multiple application teams on a shared infrastructure, we’d like to be able to charge back to these teams (or at least report) depending on usage metrics (we could use automatic tags to associate applications to individual teams).