Den Quick Start

Open In Colab

With Runhouse Den, you can manage all of your compute and make it available through ephemeral clusters for both research and production workflows. Whether you have a ML platforms team or just yourself, Runhouse is the easiest way to ensure your compute is efficiently utilized and oversubscribed without slowing down development velocity.

  • Centrally manage multiple cloud accounts and Kubernetes clusters in the same place.

  • Both developers and the platforms team can monitor resource usage in detail and access persisted logs in the web UI.

  • Save and reload services and resources (clusters, functions, classes, etc) - for instance, share an embeddings endpoint to be used by multiple teams with perfect reproducibility.

  • Set limits and quotas for teams and individuals, and queue requests for compute.

A few of the ways Runhouse Den has helped teams include:

  • Moving research teams from locally or ad hoc hosted notebooks, towards standardized, reproducible compute and training workflows that are ready to be executed in production.

  • Improving Platform team observability over day-to-day activities as the ML team grew, and getting much better disambiguation between waste (clusters up too long, clusters underutilizing their resources) and genuinely valuable workflows that just happen to be long-lived.

  • Reducing research-to-production time by 80% since resources launched and saved to Den can be scheduled nearly as-is by orchestrators.

  • Improving key / credentials management

  • Adopting a second cloud due to better quotas and cloud credits (while being mostly opaque to the end user)

If you haven’t already, check-out Quickstart or the Runhouse usage guide to see how you can get started with the Runhouse client as well.

Installing Runhouse

To use Runhouse to launch on-demand clusters, run the following installation command.

!pip install "runhouse"

Account Creation & Login

You can create an account on the run.house website or by calling the login command in Python or CLI.

To login, call runhouse login --sync-secrets in CLI. This will ask you a series of questions on whether to sync local secrets to Runhouse - e.g. your AWS / GCP / Azure secrets; once synced, you can launch compute via Runhouse anywhere you are authenticated with Runhouse.

For more information on Secrets management, refer to the Secrets Tutorial. Secrets are always securely stored in Vault. If you have any questions about Runhouse’s information security policies, please reach out at hello@run.house.

Launching and Saving Clusters

Let’s start by constructing some Runhouse resources that we’d like to save.

import runhouse as rh
cluster = rh.ondemand_cluster(
    name="rh-cluster",
    cpus="4",
    provider="aws"
).up_if_not()
def get_platform(a = 0):
        import platform
        return platform.platform()

remote_get_platform = rh.function(get_platform).to(cluster)
INFO | 2024-05-16 03:51:58.483032 | Because this function is defined in a notebook, writing it out to /Users/donny/code/notebooks/docs/get_platform_fn.py to make it importable. Please make sure the function does not rely on any local variables, including imports (which should be moved inside the function body). This restriction does not apply to functions defined in normal Python files.
INFO | 2024-05-16 03:51:58.493093 | Port 32300 is already in use. Trying next port.
INFO | 2024-05-16 03:51:58.494347 | Forwarding port 32301 to port 32300 on localhost.
INFO | 2024-05-16 03:51:59.587613 | Server rh-cluster is up.
INFO | 2024-05-16 03:51:59.595752 | Copying package from file:///Users/donny/code/notebooks to: rh-cluster
INFO | 2024-05-16 03:52:01.252665 | Sending module get_platform of type <class 'runhouse.resources.functions.function.Function'> to rh-cluster

You can save the resources we created above to your Den account with the save method:

cluster.save()
remote_get_platform.save()

Reloading

Once saved, resources can be reloaded from any environment in which you are logged into. For instance, if you are running this in a Colab notebook, you can jump into your terminal, call runhouse login, and then reconstruct and run the function on the cluster with the following Python script:

import runhouse as rh

if __name__ == "__main__":
    reloaded_fn = rh.function(name="get_platform")
    print(reloaded_fn())

The name used to reload the function is the method name by default. You can customize a function name using the following syntax:

remote_get_platform = rh.function(fn=get_platform, name="my_function").to(cluster)

Sharing

You can also share your resource with collaborators, and choose which level of access to give. Once shared, they will be able to see the resource in their dashboard as well, and be able to load and use the shared resource. They’ll need to load the resource using its full name, which includes your username (/your_username/get_platform).

remote_get_platform.share(
    users=["teammate1@email.com"],
    access_level="write",
)

Web UI

After saving your resources, you can log in and see them on your Den dashboard, labeled as /<username>/rh-cluster and /<username>/get_platform.

Clicking into the resource provides information about your resource. You can view the resource metadata, previous versions, and activity, or add a description to the resource.

Dive Deeper

Check on more in-depth tutorials on:

Previous
Quick Start
Next
How to Use Runhouse