Connect a source
Add a dataset in four steps, Source, Access, Map to the platform, and Name and schedule.
Select + Connect a source on the Datasets page. You can connect the same kind of source more than once; each becomes its own dataset.
Source
Pick where the data lives.
| Group | Source | Access |
|---|---|---|
| Files | Upload a file | CSV, Excel, or Parquet. Re-upload to refresh. |
| Files | Google Sheets | Share the sheet, or sign in with Google |
| Databases | SQL database | Postgres, MySQL, Snowflake, or BigQuery |
| Databases | Amazon RDS | An IAM role scoped to one instance |
| Object storage | Amazon S3 | An IAM role and one bucket prefix |
| Object storage | Google Cloud Storage | A service account and one bucket. Coming soon. |
| Metrics | Prometheus or Mimir | A bearer token per endpoint |
| Business systems | Salesforce | Accounts and revenue. Coming soon. |
| Business systems | Workday | Org hierarchy and headcount. Coming soon. |
Access
Grant the least access OpsLyft needs to read it. Nothing is written back.
Map to the platform
Tell OpsLyft which columns are the account, the team, and the date. This is what lets the dataset join cloud cost.
| Column role | Joins on | Example column |
|---|---|---|
| Account | Account ID or Account Name | aws_account_id |
| Team | Your owner tag or team dimension | team |
| Date | The cost date, at day or month grain | month |
Other columns become measures (such as revenue or events) and attributes (such as customer tier).
Name and schedule
Name the dataset, choose how often it refreshes, and assign an owner. The owner is notified when a sync fails.
When the first sync finishes, the dataset shows Synced and is available in the widget builder and Dimension Studio.