Set expectations for your data, implement data quality checks and data monitoring. Send alerts by email or Slack when anomalies are detected and implement your data observability.
Write a custom query that performs a check on a dataset, e.g. simply a count of the number of records:
SELECT COUNT(id) AS my_count FROM some_table
Now add an app that fetches the result of this query, and that sends out an alert in case the result is incorrect:
dbconn = pq.dbconnect('dw_123')
data = dbconn.fetch('dw_123', 'schema_name', 'table_name')
count = data[0]["my_count"]
if count[0]<10000:
slack = pq.connect("Slack") # use your name of the connection
slack.add("message", channel = "QA", text = "Data quality alert", username = "My bot")
Finally, add a schedule to your monitoring app.
Of course you can implement more advanced quality checks, based on the presense of recent timestamps, performing joins, applying a regular expression to records, a WHERE clause to select outliers etc.
Click here for more info on the Slack connector.
Watch a 2 minute demo on how to implement the above script: