GCS

Create a connection between Faraday and Google Cloud Storage so that your data is always up to date to make predictions, and your predictions can seamlessly sync back to your cloud bucket.

In this tutorial, we'll show you how to:

  • Connect your GCS account to Faraday using a connection.

Let's dive in.

  1. You'll need a Faraday Enterprise account — talk to sales to get set up.

Prerequisites

You'll need the following details to create your connection to GCS:

  • Bucket name requiredtext
  • Project ID requiredtext
  • GCP service account email optionaltextGoogle Cloud service account Faraday generated for this connection. Grant this account access on your bucket, then force-update the connection. Unique per Faraday GCS connection.

Granting access

First, you'll need Faraday access to your GCS account.

CSV is a well-known format for transferring data in large batches. Faraday can accept files up to 5gb in size. If you have more than 5gb of data, we ask that you split it into multiple files.

Faraday's CSV support is based on folders. Each CSV in a folder should have the exact same structure. There should be no more than 5000 files in a single folder.

Google Cloud Storage (GCS) can be used to send and receive files. Access is shared using Google Cloud IAM permissions. We suggest that you create a Faraday-only bucket to both send and receive data. Within this bucket, Faraday would have full read and write access.

The bucket should be dedicated to Faraday (e.g. "gcp://faraday-acme-io2quob/"). Within this bucket, Faraday should have full read and write access.

Alternatively, you can give Faraday access to certain prefixes in a shared bucket, or you can use a bucket that doesn't start with Faraday. In this case, please contact Faraday support for further assistance.

Setup process

  1. Create your Faraday connection using the API or UI. Faraday generates a unique Google Cloud service account for this connection.
  2. Copy the service account email from the connection details (gcp_service_account_email).
  3. Grant that account Storage Object Admin (roles/storage.objectAdmin) and Storage Bucket Viewer (roles/storage.bucketViewer) on the bucket. If this is a shared bucket, limit access to specific prefixes using an object prefix IAM Condition.
  4. Test the connection with Force Update. The connection will error until the grant is in place.

The same generated account is used for Datasets (send data to Faraday) and Targets (receive data from Faraday). Grant read, write, or both.

Connections created before unique identities still use Faraday's shared accounts:

  • Datasets: faraday-incoming@production-237317.iam.gserviceaccount.com
  • Targets: faraday-outgoing@production-237317.iam.gserviceaccount.com

Example prefix condition

Here is an example object prefix condition

resource.name.startsWith('projects/_/buckets/BUCKET_NAME/objects/OBJECT_PREFIX')

Faraday suggests that you use an unguessable string somewhere in the path to your data. This avoids what is called the Confused deputy problem

For example, let's say you were using S3. Instead of naming an S3 bucket s3://faraday-acme/, name it s3://faraday-acme-pwiiprz162ez. This guarantees that malicious actors cannot guess the name and request that Faraday import data from it into their account. The same logic applies to any path that is used to locate data.

Connecting

API via cURL
Dashboard

Use a POST /connections request:

curl https://api.faraday.ai/connections --json '{
  "name": "GCS",
  "options": {
    "type": "gcp-gcs-csv",
    "bucket_name": "...",
    "project_id": "..."
  }
}'
  1. Wait briefly while Faraday establishes your connection. It shouldn't take long.

Your new connection is now ready to use.