Skip to main content
Bucket connectors read Parquet files directly from your cloud storage. Evidence queries the data in place using ClickHouse’s S3 table engine—no data is copied. See Connector Types for a comparison of database vs bucket connectors.

​
How Data Updates Work

  • Row changes: When you add, update, or remove rows in a Parquet file, the changes are available immediately in Evidence
  • Schema changes: Adding or removing columns requires clicking Publish or waiting for a scheduled refresh

​
Combining Multiple Files

Multiple Parquet files in the same folder can be combined into a single source. Use the prefix option to specify the folder path. All files must share the same schema.

​
Connection Options

​
Host (my.provider.tld)
text
required
​
Bucket Name
text
required
​
Access Key
text
required
​
Secret Key
password
required
​
prefix (e.g. folder_one/2021/data)
text
​
Port
number