Feast
tl; dr; A combinator data component that installs Feast, a feature store.
Introduction
Feast is an open-source feature store. A feature store allows you to manage, govern, and trace features derived from raw data. This is useful because it helps to unify and standardise, which reduces waste, improves quality, and makes models more reproducible.
Feast does not perform any computation. You can think of it as a meta-database; a database that manages other databases. It effectively creates a cache of feature data, keyed by time. The Feast libraries and CLIs provide a consistent way of pushing or streaming new data into the cache. Downstream systems use a similar interface to access point-in-time data. Learn more about feast in the documentation.
Test Drive
The fastest way to get started is to use the test drive functionality provided by TestFaster. Click on the "Launch Test Drive" button below (opens a new window).
Launch Jupyter
Once the component has launched, click on the Jupyter link. Feast does not come with a UI. You will use Jupyter to interact with Feast via its API.
Example Notebook
Once inside Jupyter, browse to the minimal notebook, which is the official example. Follow the instructions in the notebook.
Usage
Prerequisites
Start by preparing your Kubernetes cluster using one of the infrastructure components or use your own cluster.
Component Usage
module "feast" {
source = "combinator-ml/feast/k8s"
# Optional settings go here
}
See the full configuration options below.
Requirements
No requirements.
Providers
Name | Version |
---|---|
helm | n/a |
kubernetes | n/a |
random | n/a |
Modules
No Modules.
Resources
Name |
---|
helm_release |
kubernetes_namespace |
kubernetes_secret |
random_password |
Inputs
Name | Description | Type | Default | Required |
---|---|---|---|---|
name_prefix | Prefix to be used when naming the different components of Feast | string |
"combinator" |
no |
namespace | (Optional) The namespace to install into. Defaults to feast. | string |
"feast" |
no |
Outputs
No output.