Hello,
we have just start with data exports and we have a concern with delayed events.
The order_key in the table is event_time, as I can read in the docs this is the difference between server_received_time and client_upload_time, but I cannot understand how it solves the case when the export needs to loads in Redshift delayed data, when the new data has been loaded.
We are worried about using event_time could end on a loss of data in Redshift in case of a conflict between new event just received and old event that needs to be delivered, that Redshift will reject due to the order_key. A practical example could help us in understand the process.
We would like to have an upload_timestamp as order_key in a new column with a timestamp filled when the event is loaded in Redshift instead of Amplitude. We tried to add it to the table but the export failed because of the table structure. But in case we can be sure that event_time will solves it, this column could be not necessary.
We have been thinking on using S3 export as a more flexible way to export the events.
Thank you.
Best answer by Denis Holmes
View original